[ { "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0-0_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 588, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 30, "500": 5 }, "solvent": { "CDCl3": 98, "not specified": 18, "CD3OD": 31, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unknown": 6, "null": 1, "unspecified": 2, "CDC13": 1, "Not specified": 3, "N/A": 2, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 431, "frequency": { "50": 2, "75": 5, "80": 2, "100": 26, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 28, "CDCl3": 78, "unknown": 4, "unspecified": 2, "CD3COCD3": 1, "Not specified": 2, "N/A": 2, "pyridine-d5": 7, "MeOH": 6, "not specified": 10, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1860 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 21 }, { "number": 400, "counter": 12 }, { "number": 50, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 15, "counter": 3 }, { "number": 28, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 10, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 19, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 48, "counter": 1 }, { "number": 60, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 453.2349, "counter": 1 }, { "number": 453.2352, "counter": 1 }, { "number": 7.91, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 165 }, { "number": 1, "counter": 87 }, { "number": 14, "counter": 51 }, { "number": 10, "counter": 42 }, { "number": 8, "counter": 37 }, { "number": 16, "counter": 37 }, { "number": 18, "counter": 37 }, { "number": 15, "counter": 36 }, { "number": 12, "counter": 34 }, { "number": 4, "counter": 33 }, { "number": 5, "counter": 31 }, { "number": 6, "counter": 31 }, { "number": 9, "counter": 30 }, { "number": 17, "counter": 30 }, { "number": 2, "counter": 28 }, { "number": 3, "counter": 26 }, { "number": 24, "counter": 26 }, { "number": 7, "counter": 25 }, { "number": 11, "counter": 21 }, { "number": 23, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 20, "counter": 18 }, { "number": 21, "counter": 16 }, { "number": 28, "counter": 16 }, { "number": 224, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": 27, "counter": 12 }, { "number": 211, "counter": 12 }, { "number": -10, "counter": 12 }, { "number": -9, "counter": 12 }, { "number": 239, "counter": 11 }, { "number": -7, "counter": 11 }, { "number": 100, "counter": 10 }, { "number": 196, "counter": 10 }, { "number": -5, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": -8, "counter": 9 }, { "number": -6, "counter": 8 }, { "number": -3, "counter": 8 }, { "number": 36, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": -2, "counter": 7 }, { "number": 0, "counter": 6 } ], "valid": 527, "invalid": 115, "ratioPresent": 0.9185433135935938, "nbUnion": 35395, "duplicates": 211, "invented": 113 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 481 }, "info": { "protonNMR": { "present": 589, "notPresent": 53, "valid": 229, "notValid": 360 }, "carbonNMR": { "present": 432, "notPresent": 210, "valid": 164, "notValid": 268 }, "elementalAnalysis": { "present": 263, "notPresent": 379, "valid": 213, "notValid": 50 }, "hrms": { "present": 83, "notPresent": 559, "valid": 13, "notValid": 70 } }, "logs": { "info": 73, "warn": 203, "error": 205, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 445676, "totalTokens": 965427, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.411103 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v1" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v101.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v101" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 1" }, "stats": { "nmr": { "proton": { "nbSpectra": 594, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 26, "500": 5 }, "solvent": { "CDCl3": 96, "not specified": 11, "not provided": 1, "CD3OD": 32, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 7, "null": 2, "CDC13": 1, "N/A": 1, "Not specified": 3, "unknown": 1, "pyridine-d5": 7, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 431, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "100.6": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "not specified": 7, "CD3OD": 29, "CDCl3": 76, "MeOH": 5, "null": 1, "CD3COCD3": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "TBD": 1, "acetone-d6": 4 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2183 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 15 }, { "number": 400, "counter": 8 }, { "number": 0, "counter": 3 }, { "number": 28, "counter": 3 }, { "number": 29, "counter": 3 }, { "number": 15, "counter": 2 }, { "number": 27, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": 37, "counter": 2 }, { "number": 39, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": 56, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 19, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 457.2301, "counter": 1 }, { "number": 457.2303, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.14, "counter": 1 }, { "number": 8.25, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 166 }, { "number": 1, "counter": 91 }, { "number": 14, "counter": 48 }, { "number": 10, "counter": 42 }, { "number": 16, "counter": 42 }, { "number": 6, "counter": 39 }, { "number": 15, "counter": 39 }, { "number": 9, "counter": 36 }, { "number": 8, "counter": 35 }, { "number": 12, "counter": 32 }, { "number": 2, "counter": 31 }, { "number": 4, "counter": 31 }, { "number": 5, "counter": 31 }, { "number": 18, "counter": 31 }, { "number": 3, "counter": 29 }, { "number": 17, "counter": 26 }, { "number": 11, "counter": 23 }, { "number": 24, "counter": 23 }, { "number": 20, "counter": 22 }, { "number": 21, "counter": 21 }, { "number": 23, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 28, "counter": 15 }, { "number": 22, "counter": 12 }, { "number": 27, "counter": 11 }, { "number": -8, "counter": 11 }, { "number": -5, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 100, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": 267, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": -6, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 211, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": -3, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": -2, "counter": 7 } ], "valid": 521, "invalid": 121, "ratioPresent": 0.917568296336272, "nbUnion": 35380, "duplicates": 234, "invented": 106 }, "meta": { "withName": 636, "withoutName": 6, "withMolfile": 481 }, "info": { "protonNMR": { "present": 595, "notPresent": 47, "valid": 237, "notValid": 358 }, "carbonNMR": { "present": 431, "notPresent": 211, "valid": 174, "notValid": 257 }, "elementalAnalysis": { "present": 262, "notPresent": 380, "valid": 215, "notValid": 47 }, "hrms": { "present": 82, "notPresent": 560, "valid": 12, "notValid": 70 } }, "logs": { "info": 79, "warn": 195, "error": 207, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448574, "totalTokens": 968325, "nbPrompts": 642, "version": "v101", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.416899 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v101" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v102.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v102" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 2" }, "stats": { "nmr": { "proton": { "nbSpectra": 593, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 25, "500": 6, "Not specified": 1 }, "solvent": { "CDCl3": 97, "not specified": 15, "uncertain": 1, "unspecified": 3, "CD3OD": 34, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "N/A": 3, "Not specified": 3, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "unknown": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 442, "frequency": { "50": 2, "75": 5, "80": 2, "100": 24, "125": 3, "200": 2, "Not specified": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "uncertain": 1, "unspecified": 3, "CD3OD": 31, "CDCl3": 77, "CD3COCD3": 1, "N/A": 3, "Not specified": 3, "pyridine-d5": 7, "MeOH": 4, "not specified": 8, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 637, "notParsable": 5, "empty": 1852 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 400, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 30, "counter": 6 }, { "number": 15, "counter": 4 }, { "number": 41, "counter": 3 }, { "number": 2.99, "counter": 3 }, { "number": 14, "counter": 2 }, { "number": 18, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 72, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 431 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 93 }, { "number": 14, "counter": 48 }, { "number": 8, "counter": 41 }, { "number": 16, "counter": 41 }, { "number": 10, "counter": 38 }, { "number": 5, "counter": 37 }, { "number": 18, "counter": 37 }, { "number": 6, "counter": 36 }, { "number": 9, "counter": 36 }, { "number": 12, "counter": 35 }, { "number": 15, "counter": 34 }, { "number": 3, "counter": 33 }, { "number": 2, "counter": 31 }, { "number": 4, "counter": 29 }, { "number": 17, "counter": 28 }, { "number": 7, "counter": 25 }, { "number": 24, "counter": 25 }, { "number": 21, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 20, "counter": 18 }, { "number": 11, "counter": 16 }, { "number": 22, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 224, "counter": 10 }, { "number": -8, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 100, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": -9, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": -5, "counter": 8 }, { "number": -2, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 30, "counter": 7 }, { "number": 211, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": 239, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": -10, "counter": 7 }, { "number": 70, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": 0, "counter": 5 } ], "valid": 514, "invalid": 128, "ratioPresent": 0.9158207090405271, "nbUnion": 35410, "duplicates": 200, "invented": 136 }, "meta": { "withName": 634, "withoutName": 8, "withMolfile": 476 }, "info": { "protonNMR": { "present": 595, "notPresent": 47, "valid": 231, "notValid": 364 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 170, "notValid": 273 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 215, "notValid": 46 }, "hrms": { "present": 81, "notPresent": 561, "valid": 14, "notValid": 67 } }, "logs": { "info": 80, "warn": 189, "error": 207, "fatal": 166 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447284, "totalTokens": 967035, "nbPrompts": 642, "version": "v102", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.4143189999999999 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v102" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v103.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v103" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 3" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 5, "Not specified": 1 }, "solvent": { "CDCl3": 93, "not specified": 8, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unknown": 4, "CDC13": 1, "Not specified": 2, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 440, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "Not specified": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 77, "CD3OD": 30, "unknown": 2, "CD3COCD3": 1, "Not specified": 2, "pyridine-d5": 7, "not specified": 3, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 1992 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 400, "counter": 5 }, { "number": 15, "counter": 4 }, { "number": 50, "counter": 3 }, { "number": 0, "counter": 2 }, { "number": 22, "counter": 2 }, { "number": 38, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 11, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1600, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3400, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 4.05, "counter": 1 }, { "number": 23.22, "counter": 1 }, { "number": 72.73, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 426 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 93 }, { "number": 14, "counter": 45 }, { "number": 10, "counter": 41 }, { "number": 16, "counter": 41 }, { "number": 18, "counter": 37 }, { "number": 5, "counter": 35 }, { "number": 12, "counter": 34 }, { "number": 15, "counter": 34 }, { "number": 9, "counter": 33 }, { "number": 3, "counter": 32 }, { "number": 8, "counter": 32 }, { "number": 2, "counter": 30 }, { "number": 4, "counter": 30 }, { "number": 6, "counter": 29 }, { "number": 17, "counter": 29 }, { "number": 11, "counter": 22 }, { "number": 24, "counter": 21 }, { "number": 19, "counter": 19 }, { "number": 7, "counter": 18 }, { "number": 23, "counter": 18 }, { "number": 25, "counter": 18 }, { "number": 20, "counter": 16 }, { "number": 21, "counter": 16 }, { "number": 100, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": -10, "counter": 13 }, { "number": -9, "counter": 13 }, { "number": 26, "counter": 12 }, { "number": 22, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": -4, "counter": 11 }, { "number": -7, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 30, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 33, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 196, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -8, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 31, "counter": 5 }, { "number": 32, "counter": 5 } ], "valid": 537, "invalid": 105, "ratioPresent": 0.9183763526441786, "nbUnion": 35356, "duplicates": 261, "invented": 82 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 482 }, "info": { "protonNMR": { "present": 596, "notPresent": 46, "valid": 236, "notValid": 360 }, "carbonNMR": { "present": 440, "notPresent": 202, "valid": 174, "notValid": 266 }, "elementalAnalysis": { "present": 258, "notPresent": 384, "valid": 218, "notValid": 40 }, "hrms": { "present": 80, "notPresent": 562, "valid": 14, "notValid": 66 } }, "logs": { "info": 80, "warn": 204, "error": 198, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 446787, "totalTokens": 966538, "nbPrompts": 642, "version": "v103", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.413325 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v103" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v104.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v104" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 4" }, "stats": { "nmr": { "proton": { "nbSpectra": 597, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 3, "not specified": 2 }, "solvent": { "CDCl3": 93, "not specified": 7, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 8, "null": 1, "unknown": 2, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "Not specified": 3, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 443, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CD3OD": 30, "CDCl3": 74, "MeOH": 7, "null": 1, "unknown": 1, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "not specified": 2, "Not specified": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 2187 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 400, "counter": 5 }, { "number": 0, "counter": 3 }, { "number": 15, "counter": 3 }, { "number": 29, "counter": 2 }, { "number": 43, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": 72, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 7, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 60, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 9.67, "counter": 1 }, { "number": 9.72, "counter": 1 }, { "number": 18.23, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 165 }, { "number": 1, "counter": 92 }, { "number": 14, "counter": 52 }, { "number": 10, "counter": 42 }, { "number": 8, "counter": 39 }, { "number": 12, "counter": 38 }, { "number": 16, "counter": 38 }, { "number": 15, "counter": 37 }, { "number": 9, "counter": 34 }, { "number": 3, "counter": 33 }, { "number": 6, "counter": 32 }, { "number": 18, "counter": 31 }, { "number": 2, "counter": 30 }, { "number": 4, "counter": 29 }, { "number": 17, "counter": 29 }, { "number": 5, "counter": 28 }, { "number": 24, "counter": 25 }, { "number": 25, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 7, "counter": 18 }, { "number": 20, "counter": 18 }, { "number": 19, "counter": 16 }, { "number": 23, "counter": 16 }, { "number": 28, "counter": 16 }, { "number": 22, "counter": 13 }, { "number": -10, "counter": 12 }, { "number": 21, "counter": 11 }, { "number": 100, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": -9, "counter": 11 }, { "number": -8, "counter": 11 }, { "number": -7, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": 282, "counter": 9 }, { "number": -13, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": -14, "counter": 8 }, { "number": -11, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": -12, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -2, "counter": 7 }, { "number": -6, "counter": 7 } ], "valid": 528, "invalid": 114, "ratioPresent": 0.9187809904110467, "nbUnion": 35391, "duplicates": 218, "invented": 116 }, "meta": { "withName": 637, "withoutName": 5, "withMolfile": 482 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 231, "notValid": 366 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 177, "notValid": 266 }, "elementalAnalysis": { "present": 265, "notPresent": 377, "valid": 214, "notValid": 51 }, "hrms": { "present": 88, "notPresent": 554, "valid": 13, "notValid": 75 } }, "logs": { "info": 79, "warn": 197, "error": 206, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450785, "totalTokens": 970536, "nbPrompts": 642, "version": "v104", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.421321 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v104" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v105.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v105" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 5 }, "solvent": { "CDCl3": 94, "not specified": 9, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 5, "unknown": 2, "null": 1, "CDC13": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 442, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 31, "unknown": 2, "null": 1, "CD3COCD3": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "not specified": 5, "MeOH": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2278 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 15 }, { "number": 0, "counter": 6 }, { "number": 400, "counter": 4 }, { "number": 15, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 7, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 2.66, "counter": 1 }, { "number": 8.14, "counter": 1 }, { "number": 16.74, "counter": 1 }, { "number": 72.46, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.87, "counter": 1 }, { "number": 7.96, "counter": 1 }, { "number": 15.52, "counter": 1 }, { "number": 71.65, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 432 }, { "number": -1, "counter": 161 }, { "number": 1, "counter": 91 }, { "number": 10, "counter": 42 }, { "number": 14, "counter": 39 }, { "number": 15, "counter": 38 }, { "number": 12, "counter": 37 }, { "number": 16, "counter": 37 }, { "number": 18, "counter": 36 }, { "number": 8, "counter": 35 }, { "number": 5, "counter": 33 }, { "number": 6, "counter": 31 }, { "number": 9, "counter": 31 }, { "number": 2, "counter": 30 }, { "number": 4, "counter": 30 }, { "number": 3, "counter": 25 }, { "number": 17, "counter": 24 }, { "number": 24, "counter": 23 }, { "number": 19, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 20, "counter": 18 }, { "number": 21, "counter": 18 }, { "number": 11, "counter": 17 }, { "number": 22, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": 224, "counter": 11 }, { "number": -9, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 211, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": 267, "counter": 10 }, { "number": -10, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 196, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": -14, "counter": 6 }, { "number": -11, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -4, "counter": 6 } ], "valid": 541, "invalid": 101, "ratioPresent": 0.922374971671199, "nbUnion": 35348, "duplicates": 216, "invented": 78 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 481 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 233, "notValid": 364 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 173, "notValid": 270 }, "elementalAnalysis": { "present": 256, "notPresent": 386, "valid": 211, "notValid": 45 }, "hrms": { "present": 77, "notPresent": 565, "valid": 13, "notValid": 64 } }, "logs": { "info": 85, "warn": 201, "error": 195, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449962, "totalTokens": 969713, "nbPrompts": 642, "version": "v105", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.4196749999999998 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v105" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v106.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v106" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 6" }, "stats": { "nmr": { "proton": { "nbSpectra": 598, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 5 }, "solvent": { "CDCl3": 97, "not specified": 11, "CD3OD": 35, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 9, "null": 1, "CDC13": 1, "N/A": 2, "Not specified": 2, "pyridine-d5": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "unknown": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 441, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 78, "CD3OD": 32, "null": 1, "CD3COCD3": 1, "N/A": 2, "Not specified": 2, "pyridine-d5": 7, "MeOH": 5, "not specified": 8, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "Not provided": 1, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2266 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 24 }, { "number": 400, "counter": 5 }, { "number": 50, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 29, "counter": 3 }, { "number": -18, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1600, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3400, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 5.05, "counter": 1 }, { "number": 5.15, "counter": 1 }, { "number": 16.69, "counter": 1 }, { "number": 16.99, "counter": 1 }, { "number": 17.21, "counter": 1 }, { "number": 17.39, "counter": 1 }, { "number": 60.65, "counter": 1 }, { "number": 60.87, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 94 }, { "number": 14, "counter": 49 }, { "number": 8, "counter": 40 }, { "number": 10, "counter": 39 }, { "number": 12, "counter": 39 }, { "number": 15, "counter": 36 }, { "number": 16, "counter": 36 }, { "number": 4, "counter": 33 }, { "number": 6, "counter": 33 }, { "number": 9, "counter": 33 }, { "number": 18, "counter": 33 }, { "number": 5, "counter": 31 }, { "number": 2, "counter": 29 }, { "number": 3, "counter": 29 }, { "number": 17, "counter": 28 }, { "number": 24, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 19, "counter": 19 }, { "number": 28, "counter": 17 }, { "number": 20, "counter": 16 }, { "number": 21, "counter": 16 }, { "number": -10, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": -9, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": 224, "counter": 11 }, { "number": -8, "counter": 11 }, { "number": -7, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 211, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": 196, "counter": 8 }, { "number": -14, "counter": 8 }, { "number": -13, "counter": 8 }, { "number": -11, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -20, "counter": 7 }, { "number": -16, "counter": 7 }, { "number": -15, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 32, "counter": 6 } ], "valid": 530, "invalid": 112, "ratioPresent": 0.9224860730125408, "nbUnion": 35382, "duplicates": 265, "invented": 106 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 483 }, "info": { "protonNMR": { "present": 599, "notPresent": 43, "valid": 229, "notValid": 370 }, "carbonNMR": { "present": 442, "notPresent": 200, "valid": 172, "notValid": 270 }, "elementalAnalysis": { "present": 263, "notPresent": 379, "valid": 215, "notValid": 48 }, "hrms": { "present": 82, "notPresent": 560, "valid": 11, "notValid": 71 } }, "logs": { "info": 78, "warn": 191, "error": 214, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449411, "totalTokens": 969162, "nbPrompts": 642, "version": "v106", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.4185729999999999 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v106" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v107.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v107" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 7" }, "stats": { "nmr": { "proton": { "nbSpectra": 598, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 3, "Not specified": 1, "not specified": 1 }, "solvent": { "CDCl3": 94, "not specified": 11, "CD3OD": 33, "unknown": 5, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 5, "Not specified": 3, "MeOD": 3, "unspecified": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 442, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "Not specified": 1, "50.3": 1, "75.4": 1, "not specified": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 30, "CDCl3": 75, "null": 2, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "not specified": 8, "MeOH": 5, "Not specified": 3, "unknown": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "unspecified": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1937 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 20 }, { "number": 0, "counter": 6 }, { "number": 400, "counter": 5 }, { "number": 32, "counter": 4 }, { "number": 15, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 8.5, "counter": 2 }, { "number": 5, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 60, "counter": 1 }, { "number": 72, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 9.67, "counter": 1 }, { "number": 9.7, "counter": 1 }, { "number": 18.25, "counter": 1 }, { "number": 18.33, "counter": 1 }, { "number": 72.05, "counter": 1 }, { "number": 296.5, "counter": 1 }, { "number": 3.77, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 425 }, { "number": -1, "counter": 162 }, { "number": 1, "counter": 89 }, { "number": 14, "counter": 49 }, { "number": 10, "counter": 43 }, { "number": 15, "counter": 40 }, { "number": 16, "counter": 38 }, { "number": 9, "counter": 37 }, { "number": 18, "counter": 37 }, { "number": 8, "counter": 36 }, { "number": 12, "counter": 34 }, { "number": 6, "counter": 33 }, { "number": 5, "counter": 31 }, { "number": 2, "counter": 30 }, { "number": 3, "counter": 28 }, { "number": 4, "counter": 28 }, { "number": 17, "counter": 24 }, { "number": 25, "counter": 23 }, { "number": 24, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 23, "counter": 20 }, { "number": 7, "counter": 18 }, { "number": 20, "counter": 18 }, { "number": 100, "counter": 18 }, { "number": 19, "counter": 15 }, { "number": 21, "counter": 13 }, { "number": 28, "counter": 13 }, { "number": 224, "counter": 12 }, { "number": 22, "counter": 10 }, { "number": 26, "counter": 10 }, { "number": 27, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": 239, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": 33, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -7, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 31, "counter": 5 }, { "number": 34, "counter": 5 } ], "valid": 528, "invalid": 114, "ratioPresent": 0.9185796157815908, "nbUnion": 35384, "duplicates": 202, "invented": 112 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 483 }, "info": { "protonNMR": { "present": 600, "notPresent": 42, "valid": 231, "notValid": 369 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 171, "notValid": 272 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 217, "notValid": 47 }, "hrms": { "present": 83, "notPresent": 559, "valid": 12, "notValid": 71 } }, "logs": { "info": 75, "warn": 195, "error": 213, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448598, "totalTokens": 968349, "nbPrompts": 642, "version": "v107", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.416947 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v107" }, { "file": { "name": "gpt-3.5-turbo-1106-0-0_v108.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v108" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 8" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 19, "500": 4 }, "solvent": { "CDCl3": 95, "uncertain": 1, "not specified": 7, "CD3OD": 34, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unknown": 2, "null": 2, "CDC13": 1, "pyridine-d5": 7, "MeOH": 8, "Not specified": 1, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 439, "frequency": { "50": 2, "75": 5, "80": 2, "100": 17, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 76, "CD3OD": 31, "unknown": 2, "null": 1, "CD3COCD3": 1, "pyridine-d5": 7, "MeOH": 8, "Not specified": 1, "not specified": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 638, "notParsable": 4, "empty": 2124 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 12 }, { "number": 50, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 15, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 13, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 400, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.88, "counter": 1 }, { "number": 5.34, "counter": 1 }, { "number": 11.15, "counter": 1 }, { "number": 64.63, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 296.5, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.14, "counter": 1 }, { "number": 7.8, "counter": 1 }, { "number": 11.41, "counter": 1 }, { "number": 77.65, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": -3, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 433 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 97 }, { "number": 10, "counter": 43 }, { "number": 14, "counter": 43 }, { "number": 8, "counter": 39 }, { "number": 12, "counter": 37 }, { "number": 15, "counter": 37 }, { "number": 16, "counter": 37 }, { "number": 3, "counter": 36 }, { "number": 18, "counter": 35 }, { "number": 5, "counter": 34 }, { "number": 2, "counter": 32 }, { "number": 4, "counter": 32 }, { "number": 9, "counter": 32 }, { "number": 6, "counter": 29 }, { "number": 17, "counter": 27 }, { "number": 7, "counter": 24 }, { "number": 24, "counter": 23 }, { "number": 25, "counter": 22 }, { "number": 11, "counter": 19 }, { "number": 19, "counter": 19 }, { "number": 20, "counter": 19 }, { "number": 23, "counter": 18 }, { "number": 22, "counter": 16 }, { "number": 28, "counter": 16 }, { "number": 100, "counter": 15 }, { "number": 21, "counter": 14 }, { "number": 211, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": -10, "counter": 11 }, { "number": -9, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 70, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": 282, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": -3, "counter": 7 }, { "number": -2, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 43, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -5, "counter": 6 } ], "valid": 543, "invalid": 99, "ratioPresent": 0.9155163030361466, "nbUnion": 35344, "duplicates": 175, "invented": 70 }, "meta": { "withName": 635, "withoutName": 7, "withMolfile": 479 }, "info": { "protonNMR": { "present": 598, "notPresent": 44, "valid": 234, "notValid": 364 }, "carbonNMR": { "present": 441, "notPresent": 201, "valid": 174, "notValid": 267 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 217, "notValid": 44 }, "hrms": { "present": 86, "notPresent": 556, "valid": 15, "notValid": 71 } }, "logs": { "info": 84, "warn": 197, "error": 198, "fatal": 163 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449258, "totalTokens": 969009, "nbPrompts": 642, "version": "v108", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.418267 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0_0", "version": "v108" } ], "versions": [ "v1", "v101", "v102", "v103", "v104", "v105", "v106", "v107", "v108" ], "stats": { "yaml": { "parsable": [ 640, 639, 637, 641, 640, 641, 642, 640, 638 ], "notParsable": [ 2, 3, 5, 1, 2, 1, 0, 2, 4 ], "empty": [ 1860, 2183, 1852, 1992, 2187, 2278, 2266, 1937, 2124 ] }, "numbers": { "valid": [ 527, 521, 514, 537, 528, 541, 530, 528, 543 ], "invalid": [ 115, 121, 128, 105, 114, 101, 112, 114, 99 ], "ratioPresent": [ 0.9185433135935938, 0.917568296336272, 0.9158207090405271, 0.9183763526441786, 0.9187809904110467, 0.922374971671199, 0.9224860730125408, 0.9185796157815908, 0.9155163030361466 ], "nbUnion": [ 35395, 35380, 35410, 35356, 35391, 35348, 35382, 35384, 35344 ], "duplicates": [ 211, 234, 200, 261, 218, 216, 265, 202, 175 ], "invented": [ 113, 106, 136, 82, 116, 78, 106, 112, 70 ] }, "meta": { "withName": [ 638, 636, 634, 638, 637, 638, 639, 639, 635 ], "withoutName": [ 4, 6, 8, 4, 5, 4, 3, 3, 7 ], "withMolfile": [ 481, 481, 476, 482, 482, 481, 483, 483, 479 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0-0_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 563, "frequency": { "50": 4, "80": 2, "200": 3, "300": 11, "400": 43, "500": 7, "1.21": 1 }, "solvent": { "unknown": 98, "CDCl3": 96, "not specified": 14, "CD3OD": 28, "CHCl3": 2, "DMSO-d6": 71, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "N/A": 3, "pyridine-d5": 6, "δ": 1, "MeOH": 4, "methanol": 1, "ppm": 1, "MeOD": 3, "Not specified": 2, "δ/ppm": 1, "D2O": 3, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "CHCl3-EtOH (1:1)": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 471, "frequency": { "50": 2, "75": 5, "80": 2, "100": 43, "125": 3, "200": 1, "50.3": 1, "75.4": 1 }, "solvent": { "unknown": 81, "DMSO-d6": 52, "CDCl3": 87, "not specified": 14, "CD3OD": 26, "CHCl3": 2, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 6, "δ": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "Not specified": 2, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "CHCl3-EtOH (1:1)": 1, "acetone-d6": 5, "MeOH": 1 } } }, "yaml": { "parsable": 621, "notParsable": 21, "empty": 15422 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 161 }, { "number": 400, "counter": 28 }, { "number": 100, "counter": 8 }, { "number": -18, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 3, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 148 }, { "number": 1, "counter": 80 }, { "number": 14, "counter": 50 }, { "number": 100, "counter": 47 }, { "number": 4, "counter": 46 }, { "number": 10, "counter": 45 }, { "number": 2, "counter": 44 }, { "number": 3, "counter": 43 }, { "number": 5, "counter": 43 }, { "number": 15, "counter": 40 }, { "number": 8, "counter": 39 }, { "number": 6, "counter": 37 }, { "number": 16, "counter": 35 }, { "number": 17, "counter": 32 }, { "number": 12, "counter": 31 }, { "number": 9, "counter": 30 }, { "number": 11, "counter": 27 }, { "number": 18, "counter": 27 }, { "number": 24, "counter": 25 }, { "number": 7, "counter": 24 }, { "number": 25, "counter": 23 }, { "number": 20, "counter": 19 }, { "number": 21, "counter": 19 }, { "number": -3, "counter": 19 }, { "number": -8, "counter": 19 }, { "number": 30, "counter": 18 }, { "number": 211, "counter": 18 }, { "number": 224, "counter": 18 }, { "number": 19, "counter": 17 }, { "number": 23, "counter": 17 }, { "number": 239, "counter": 17 }, { "number": 196, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": -2, "counter": 16 }, { "number": 43, "counter": 15 }, { "number": -13, "counter": 15 }, { "number": -5, "counter": 15 }, { "number": 22, "counter": 14 }, { "number": 27, "counter": 14 }, { "number": 28, "counter": 14 }, { "number": -7, "counter": 14 }, { "number": 26, "counter": 13 }, { "number": 70, "counter": 13 }, { "number": -10, "counter": 13 }, { "number": 254, "counter": 12 }, { "number": -12, "counter": 12 }, { "number": 31, "counter": 11 }, { "number": 135, "counter": 11 }, { "number": -6, "counter": 11 } ], "valid": 370, "invalid": 272, "ratioPresent": 0.883372568806167, "nbUnion": 35508, "duplicates": 304, "invented": 638 }, "meta": { "withName": 613, "withoutName": 29, "withMolfile": 469 }, "info": { "protonNMR": { "present": 584, "notPresent": 58, "valid": 166, "notValid": 418 }, "carbonNMR": { "present": 509, "notPresent": 133, "valid": 142, "notValid": 367 }, "elementalAnalysis": { "present": 267, "notPresent": 375, "valid": 222, "notValid": 45 }, "hrms": { "present": 53, "notPresent": 589, "valid": 10, "notValid": 43 } }, "logs": { "info": 30, "warn": 184, "error": 255, "fatal": 173 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 529948, "totalTokens": 1016957, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.546905 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0_0", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 621 ], "notParsable": [ 21 ], "empty": [ 15422 ] }, "numbers": { "valid": [ 370 ], "invalid": [ 272 ], "ratioPresent": [ 0.883372568806167 ], "nbUnion": [ 35508 ], "duplicates": [ 304 ], "invented": [ 638 ] }, "meta": { "withName": [ 613 ], "withoutName": [ 29 ], "withMolfile": [ 469 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_0_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0-0_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 580, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 29, "500": 5 }, "solvent": { "not specified": 10, "N/A": 7, "DMSO-d6": 71, "unknown": 6, "CDCl3": 79, "CD3OD": 33, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 6, "MeOH": 9, "Not specified": 1, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 2, "acetone-d6": 3, "mixture": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 432, "frequency": { "50": 2, "75": 5, "80": 2, "100": 26, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1, "174.9": 1, "100.6": 1 }, "solvent": { "DMSO-d6": 50, "unknown": 6, "not specified": 1, "CD3OD": 31, "CHCl3": 2, "CDCl3": 66, "CD3COCD3": 1, "N/A": 5, "pyridine-d5": 6, "MeOH": 6, "Not specified": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "ppm": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 625, "notParsable": 17, "empty": 2660 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 0, "counter": 12 }, { "number": 400, "counter": 11 }, { "number": -18, "counter": 2 }, { "number": 15, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": -3, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 432 }, { "number": -1, "counter": 169 }, { "number": 1, "counter": 79 }, { "number": 14, "counter": 64 }, { "number": 10, "counter": 43 }, { "number": 4, "counter": 42 }, { "number": 3, "counter": 40 }, { "number": 6, "counter": 40 }, { "number": 18, "counter": 40 }, { "number": 15, "counter": 39 }, { "number": 16, "counter": 37 }, { "number": 12, "counter": 36 }, { "number": 5, "counter": 35 }, { "number": 2, "counter": 34 }, { "number": 9, "counter": 33 }, { "number": 8, "counter": 31 }, { "number": 17, "counter": 29 }, { "number": 20, "counter": 27 }, { "number": 23, "counter": 25 }, { "number": 24, "counter": 25 }, { "number": -10, "counter": 25 }, { "number": 11, "counter": 24 }, { "number": 19, "counter": 24 }, { "number": 25, "counter": 23 }, { "number": -7, "counter": 23 }, { "number": 21, "counter": 22 }, { "number": 7, "counter": 21 }, { "number": -8, "counter": 20 }, { "number": 22, "counter": 17 }, { "number": -13, "counter": 17 }, { "number": 28, "counter": 14 }, { "number": -9, "counter": 13 }, { "number": -6, "counter": 12 }, { "number": 100, "counter": 11 }, { "number": -4, "counter": 11 }, { "number": -3, "counter": 11 }, { "number": -12, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": -2, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": -11, "counter": 8 }, { "number": 33, "counter": 7 }, { "number": -19, "counter": 7 }, { "number": -15, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": 36, "counter": 6 }, { "number": 8.2, "counter": 6 }, { "number": 4.2, "counter": 6 }, { "number": 7.3, "counter": 6 } ], "valid": 512, "invalid": 130, "ratioPresent": 0.9015354624557821, "nbUnion": 35350, "duplicates": 337, "invented": 89 }, "meta": { "withName": 609, "withoutName": 33, "withMolfile": 455 }, "info": { "protonNMR": { "present": 582, "notPresent": 60, "valid": 213, "notValid": 369 }, "carbonNMR": { "present": 433, "notPresent": 209, "valid": 169, "notValid": 264 }, "elementalAnalysis": { "present": 250, "notPresent": 392, "valid": 207, "notValid": 43 }, "hrms": { "present": 86, "notPresent": 556, "valid": 14, "notValid": 72 } }, "logs": { "info": 64, "warn": 210, "error": 181, "fatal": 187 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 468008, "totalTokens": 944745, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.412753 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_0_0", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 625 ], "notParsable": [ 17 ], "empty": [ 2660 ] }, "numbers": { "valid": [ 512 ], "invalid": [ 130 ], "ratioPresent": [ 0.9015354624557821 ], "nbUnion": [ 35350 ], "duplicates": [ 337 ], "invented": [ 89 ] }, "meta": { "withName": [ 609 ], "withoutName": [ 33 ], "withMolfile": [ 455 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_0_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0-0_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 562, "frequency": { "20": 1, "50": 4, "80": 2, "200": 4, "300": 8, "400": 37, "500": 3, "Not specified": 1, "12.62": 1 }, "solvent": { "not specified": 44, "DMSO-d6": 67, "unknown": 5, "CDCl3": 89, "CD3OD": 31, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 9, "CDC13": 1, "Not specified": 6, "pyridine-d5": 7, "MeOD": 2, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "δ": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 444, "frequency": { "20": 1, "50": 2, "75": 4, "80": 2, "100": 33, "125": 3, "200": 2, "400": 1, "75.4": 1, "100.6": 1, "Not specified": 1, "156.2": 1 }, "solvent": { "DMSO-d6": 49, "not specified": 29, "CD3OD": 30, "CHCl3": 2, "CDCl3": 76, "MeOH": 8, "CD3COCD3": 1, "Not specified": 5, "unknown": 2, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 2, "DMSO": 2, "D2O": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "δ": 1, "CH3OH": 1 } } }, "yaml": { "parsable": 605, "notParsable": 37, "empty": 1900 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 38 }, { "number": 400, "counter": 22 }, { "number": 50, "counter": 8 }, { "number": 0, "counter": 7 }, { "number": 6, "counter": 2 }, { "number": 11, "counter": 2 }, { "number": 12, "counter": 2 }, { "number": 15, "counter": 2 }, { "number": 33, "counter": 2 }, { "number": 43, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 7, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 10, "counter": 1 }, { "number": 13, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 16, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 7.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 433 }, { "number": -1, "counter": 183 }, { "number": 1, "counter": 139 }, { "number": 2, "counter": 63 }, { "number": 3, "counter": 61 }, { "number": 8, "counter": 57 }, { "number": 6, "counter": 56 }, { "number": 4, "counter": 55 }, { "number": 5, "counter": 52 }, { "number": 10, "counter": 39 }, { "number": 14, "counter": 39 }, { "number": 9, "counter": 38 }, { "number": 7, "counter": 33 }, { "number": 100, "counter": 31 }, { "number": 18, "counter": 28 }, { "number": 15, "counter": 27 }, { "number": 12, "counter": 26 }, { "number": 17, "counter": 23 }, { "number": 16, "counter": 22 }, { "number": -2, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 25, "counter": 21 }, { "number": 23, "counter": 20 }, { "number": -3, "counter": 20 }, { "number": 24, "counter": 18 }, { "number": 19, "counter": 17 }, { "number": -4, "counter": 16 }, { "number": 21, "counter": 14 }, { "number": 28, "counter": 12 }, { "number": 20, "counter": 11 }, { "number": 22, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": -5, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": 30, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 50, "counter": 7 }, { "number": 63, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 91, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": 90, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 8.2, "counter": 6 }, { "number": 8.6, "counter": 6 }, { "number": 6.6, "counter": 6 } ], "valid": 472, "invalid": 170, "ratioPresent": 0.8803162582651696, "nbUnion": 35409, "duplicates": 330, "invented": 140 }, "meta": { "withName": 587, "withoutName": 55, "withMolfile": 444 }, "info": { "protonNMR": { "present": 564, "notPresent": 78, "valid": 200, "notValid": 364 }, "carbonNMR": { "present": 444, "notPresent": 198, "valid": 159, "notValid": 285 }, "elementalAnalysis": { "present": 234, "notPresent": 408, "valid": 199, "notValid": 35 }, "hrms": { "present": 101, "notPresent": 541, "valid": 10, "notValid": 91 } }, "logs": { "info": 63, "warn": 146, "error": 235, "fatal": 198 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 464653, "totalTokens": 938822, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.403475 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_0_0", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 605 ], "notParsable": [ 37 ], "empty": [ 1900 ] }, "numbers": { "valid": [ 472 ], "invalid": [ 170 ], "ratioPresent": [ 0.8803162582651696 ], "nbUnion": [ 35409 ], "duplicates": [ 330 ], "invented": [ 140 ] }, "meta": { "withName": [ 587 ], "withoutName": [ 55 ], "withMolfile": [ 444 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_0_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0-0_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0-0_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 591, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 4 }, "solvent": { "CDCl3": 86, "Not specified": 1, "DMSO-d6": 74, "CD3OD": 32, "unknown": 1, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 13, "CDC13": 1, "N/A": 1, "pyridine-d5": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 436, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 29, "CHCl3": 1, "CDCl3": 68, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 7, "MeOH": 8, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 1651 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 12 }, { "number": 100, "counter": 10 }, { "number": 400, "counter": 6 }, { "number": -18, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 433 }, { "number": -1, "counter": 172 }, { "number": 1, "counter": 74 }, { "number": 14, "counter": 46 }, { "number": 4, "counter": 43 }, { "number": 10, "counter": 43 }, { "number": 15, "counter": 42 }, { "number": 16, "counter": 39 }, { "number": 5, "counter": 38 }, { "number": 18, "counter": 38 }, { "number": 8, "counter": 36 }, { "number": 3, "counter": 35 }, { "number": 6, "counter": 34 }, { "number": 9, "counter": 34 }, { "number": 12, "counter": 33 }, { "number": 17, "counter": 28 }, { "number": 2, "counter": 26 }, { "number": 24, "counter": 25 }, { "number": 7, "counter": 22 }, { "number": 11, "counter": 22 }, { "number": 25, "counter": 21 }, { "number": 19, "counter": 20 }, { "number": 21, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 28, "counter": 18 }, { "number": 20, "counter": 16 }, { "number": -10, "counter": 16 }, { "number": 22, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": -7, "counter": 12 }, { "number": -9, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 30, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": -8, "counter": 9 }, { "number": -3, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 211, "counter": 8 }, { "number": 239, "counter": 7 }, { "number": -13, "counter": 7 }, { "number": 70, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": 29, "counter": 5 }, { "number": 32, "counter": 5 }, { "number": 33, "counter": 5 }, { "number": 36, "counter": 5 }, { "number": 230, "counter": 5 } ], "valid": 522, "invalid": 120, "ratioPresent": 0.9185033261071314, "nbUnion": 35336, "duplicates": 238, "invented": 128 }, "meta": { "withName": 620, "withoutName": 22, "withMolfile": 465 }, "info": { "protonNMR": { "present": 591, "notPresent": 51, "valid": 231, "notValid": 360 }, "carbonNMR": { "present": 436, "notPresent": 206, "valid": 173, "notValid": 263 }, "elementalAnalysis": { "present": 252, "notPresent": 390, "valid": 214, "notValid": 38 }, "hrms": { "present": 83, "notPresent": 559, "valid": 10, "notValid": 73 } }, "logs": { "info": 70, "warn": 206, "error": 189, "fatal": 177 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 444311, "totalTokens": 926826, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 0, "top_p": 0, "cost": 1.371137 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_0_0", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 641 ], "notParsable": [ 1 ], "empty": [ 1651 ] }, "numbers": { "valid": [ 522 ], "invalid": [ 120 ], "ratioPresent": [ 0.9185033261071314 ], "nbUnion": [ 35336 ], "duplicates": [ 238 ], "invented": [ 128 ] }, "meta": { "withName": [ 620 ], "withoutName": [ 22 ], "withMolfile": [ 465 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.00001_0.00001", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.00001, "top_p": 0.00001, "promptDescription": "Our baseline prompt. Replicate 3" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.00001-0.00001_v103.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.00001-0.00001_v103" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.00001, "top_p": 0.00001, "promptDescription": "Our baseline prompt. Replicate 3" }, "stats": { "nmr": { "proton": { "nbSpectra": 593, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 6 }, "solvent": { "CDCl3": 94, "not specified": 9, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 1, "unknown": 2, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 438, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "not specified": 6, "CDCl3": 78, "CD3OD": 30, "null": 2, "CD3COCD3": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "unknown": 1 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2142 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 15 }, { "number": 15, "counter": 3 }, { "number": 400, "counter": 3 }, { "number": 0, "counter": 2 }, { "number": 10, "counter": 2 }, { "number": 32, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 7, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 60, "counter": 1 }, { "number": 75, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 425 }, { "number": -1, "counter": 165 }, { "number": 1, "counter": 92 }, { "number": 14, "counter": 49 }, { "number": 16, "counter": 43 }, { "number": 10, "counter": 42 }, { "number": 15, "counter": 40 }, { "number": 3, "counter": 35 }, { "number": 8, "counter": 35 }, { "number": 9, "counter": 35 }, { "number": 12, "counter": 34 }, { "number": 4, "counter": 33 }, { "number": 18, "counter": 33 }, { "number": 2, "counter": 32 }, { "number": 6, "counter": 31 }, { "number": 5, "counter": 28 }, { "number": 11, "counter": 24 }, { "number": 24, "counter": 24 }, { "number": 17, "counter": 23 }, { "number": 7, "counter": 21 }, { "number": 25, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 20, "counter": 19 }, { "number": 23, "counter": 18 }, { "number": 21, "counter": 17 }, { "number": 22, "counter": 14 }, { "number": 28, "counter": 14 }, { "number": 100, "counter": 11 }, { "number": 224, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 239, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 30, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 31, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 36, "counter": 5 }, { "number": 48, "counter": 5 }, { "number": 254, "counter": 5 }, { "number": 282, "counter": 5 } ], "valid": 536, "invalid": 106, "ratioPresent": 0.920583860289208, "nbUnion": 35366, "duplicates": 187, "invented": 92 }, "meta": { "withName": 640, "withoutName": 2, "withMolfile": 484 }, "info": { "protonNMR": { "present": 593, "notPresent": 49, "valid": 238, "notValid": 355 }, "carbonNMR": { "present": 439, "notPresent": 203, "valid": 174, "notValid": 265 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 214, "notValid": 47 }, "hrms": { "present": 85, "notPresent": 557, "valid": 14, "notValid": 71 } }, "logs": { "info": 77, "warn": 201, "error": 206, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447679, "totalTokens": 967430, "nbPrompts": 642, "version": "v103", "model": "gpt-3.5-turbo-1106", "temperature": 0.00001, "top_p": 0.00001, "cost": 1.415109 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.00001_0.00001", "version": "v103" } ], "versions": [ "v103" ], "stats": { "yaml": { "parsable": [ 641 ], "notParsable": [ 1 ], "empty": [ 2142 ] }, "numbers": { "valid": [ 536 ], "invalid": [ 106 ], "ratioPresent": [ 0.920583860289208 ], "nbUnion": [ 35366 ], "duplicates": [ 187 ], "invented": [ 92 ] }, "meta": { "withName": [ 640 ], "withoutName": [ 2 ], "withMolfile": [ 484 ] } }, "description": "Our baseline prompt" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 601, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 6 }, "solvent": { "CDCl3": 95, "not specified": 10, "uncertain": 1, "CD3OD": 34, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "Not specified": 5, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "unknown": 1, "acetone-d6": 3, "unspecified": 1 } }, "carbon": { "nbSpectra": 439, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "uncertain": 1, "CD3OD": 31, "CDCl3": 75, "null": 1, "CD3COCD3": 1, "Not specified": 4, "pyridine-d5": 7, "MeOH": 7, "not specified": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "unspecified": 1 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2015 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 12 }, { "number": 15, "counter": 5 }, { "number": 400, "counter": 5 }, { "number": 50, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 29, "counter": 3 }, { "number": 30, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2.71, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 3.73, "counter": 1 }, { "number": 3.92, "counter": 1 }, { "number": 7.46, "counter": 1 }, { "number": 7.63, "counter": 1 }, { "number": 28.11, "counter": 1 }, { "number": 28.21, "counter": 1 }, { "number": 60.41, "counter": 1 }, { "number": 60.53, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 162 }, { "number": 1, "counter": 94 }, { "number": 14, "counter": 44 }, { "number": 10, "counter": 41 }, { "number": 4, "counter": 36 }, { "number": 16, "counter": 36 }, { "number": 8, "counter": 35 }, { "number": 6, "counter": 34 }, { "number": 18, "counter": 34 }, { "number": 12, "counter": 33 }, { "number": 3, "counter": 32 }, { "number": 15, "counter": 32 }, { "number": 5, "counter": 31 }, { "number": 9, "counter": 31 }, { "number": 17, "counter": 28 }, { "number": 2, "counter": 26 }, { "number": 24, "counter": 24 }, { "number": 11, "counter": 22 }, { "number": 25, "counter": 22 }, { "number": 7, "counter": 18 }, { "number": 19, "counter": 18 }, { "number": 23, "counter": 18 }, { "number": 20, "counter": 16 }, { "number": 100, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 28, "counter": 13 }, { "number": 211, "counter": 12 }, { "number": 224, "counter": 12 }, { "number": 22, "counter": 11 }, { "number": 239, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 196, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 34, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 32, "counter": 5 } ], "valid": 524, "invalid": 118, "ratioPresent": 0.9223823368962465, "nbUnion": 35374, "duplicates": 206, "invented": 99 }, "meta": { "withName": 635, "withoutName": 7, "withMolfile": 481 }, "info": { "protonNMR": { "present": 602, "notPresent": 40, "valid": 237, "notValid": 365 }, "carbonNMR": { "present": 439, "notPresent": 203, "valid": 173, "notValid": 266 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 218, "notValid": 46 }, "hrms": { "present": 91, "notPresent": 551, "valid": 11, "notValid": 80 } }, "logs": { "info": 77, "warn": 196, "error": 208, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450940, "totalTokens": 970691, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.421631 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v1" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v101.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v101" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 1" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 6 }, "solvent": { "CDCl3": 94, "not specified": 9, "CD3OD": 35, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "N/A": 1, "literature": 1, "pyridine-d5": 7, "MeOH": 6, "Not specified": 2, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "unknown": 1 } }, "carbon": { "nbSpectra": 439, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 32, "null": 1, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 7, "MeOH": 5, "not specified": 6, "Not specified": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "unknown": 1 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2222 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 23 }, { "number": 0, "counter": 4 }, { "number": 35, "counter": 3 }, { "number": 50, "counter": 3 }, { "number": 400, "counter": 3 }, { "number": 21, "counter": 2 }, { "number": 32, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 20, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 72, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.25, "counter": 1 }, { "number": 8.14, "counter": 1 }, { "number": 10.79, "counter": 1 }, { "number": 77.82, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 392.2757, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 6.79, "counter": 1 }, { "number": 6.81, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 96 }, { "number": 14, "counter": 49 }, { "number": 8, "counter": 44 }, { "number": 10, "counter": 43 }, { "number": 4, "counter": 39 }, { "number": 2, "counter": 38 }, { "number": 6, "counter": 38 }, { "number": 12, "counter": 38 }, { "number": 16, "counter": 38 }, { "number": 18, "counter": 36 }, { "number": 15, "counter": 35 }, { "number": 9, "counter": 34 }, { "number": 5, "counter": 32 }, { "number": 17, "counter": 32 }, { "number": 3, "counter": 31 }, { "number": 11, "counter": 23 }, { "number": 7, "counter": 22 }, { "number": 24, "counter": 21 }, { "number": 23, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 20, "counter": 17 }, { "number": 28, "counter": 16 }, { "number": 22, "counter": 15 }, { "number": 100, "counter": 15 }, { "number": 21, "counter": 13 }, { "number": 211, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": 239, "counter": 11 }, { "number": -10, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 27, "counter": 9 }, { "number": 196, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": -8, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": 282, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -13, "counter": 6 }, { "number": 0, "counter": 5 } ], "valid": 526, "invalid": 116, "ratioPresent": 0.9152301112083121, "nbUnion": 35369, "duplicates": 248, "invented": 86 }, "meta": { "withName": 637, "withoutName": 5, "withMolfile": 481 }, "info": { "protonNMR": { "present": 598, "notPresent": 44, "valid": 236, "notValid": 362 }, "carbonNMR": { "present": 441, "notPresent": 201, "valid": 170, "notValid": 271 }, "elementalAnalysis": { "present": 262, "notPresent": 380, "valid": 216, "notValid": 46 }, "hrms": { "present": 91, "notPresent": 551, "valid": 14, "notValid": 77 } }, "logs": { "info": 72, "warn": 188, "error": 221, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 446626, "totalTokens": 966377, "nbPrompts": 642, "version": "v101", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.4130029999999998 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v101" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v102.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v102" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 2." }, "stats": { "nmr": { "proton": { "nbSpectra": 595, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 6 }, "solvent": { "CDCl3": 95, "not specified": 10, "uncertain": 1, "CD3OD": 34, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unspecified": 1, "null": 1, "CDC13": 1, "N/A": 2, "Not specified": 4, "unknown": 4, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 440, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 31, "unspecified": 1, "null": 1, "CD3COCD3": 1, "N/A": 2, "Not specified": 4, "pyridine-d5": 7, "MeOH": 6, "not specified": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "unknown": 1 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2064 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 16 }, { "number": 0, "counter": 4 }, { "number": 400, "counter": 3 }, { "number": 15, "counter": 2 }, { "number": 24, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 2.44, "counter": 1 }, { "number": 2.7, "counter": 1 }, { "number": 8.17, "counter": 1 }, { "number": 8.29, "counter": 1 }, { "number": 28.36, "counter": 1 }, { "number": 28.52, "counter": 1 }, { "number": 60.65, "counter": 1 }, { "number": 60.87, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 165 }, { "number": 1, "counter": 100 }, { "number": 14, "counter": 46 }, { "number": 10, "counter": 44 }, { "number": 15, "counter": 41 }, { "number": 12, "counter": 40 }, { "number": 16, "counter": 38 }, { "number": 4, "counter": 36 }, { "number": 8, "counter": 36 }, { "number": 6, "counter": 33 }, { "number": 9, "counter": 33 }, { "number": 5, "counter": 32 }, { "number": 18, "counter": 32 }, { "number": 3, "counter": 31 }, { "number": 2, "counter": 30 }, { "number": 17, "counter": 25 }, { "number": 7, "counter": 21 }, { "number": 11, "counter": 21 }, { "number": 24, "counter": 20 }, { "number": 23, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 20, "counter": 18 }, { "number": 21, "counter": 17 }, { "number": 28, "counter": 15 }, { "number": 22, "counter": 13 }, { "number": 100, "counter": 11 }, { "number": 211, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": -10, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 33, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -7, "counter": 7 }, { "number": 70, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 5 }, { "number": 32, "counter": 5 }, { "number": 230, "counter": 5 }, { "number": 240, "counter": 5 }, { "number": 0.1, "counter": 5 } ], "valid": 534, "invalid": 108, "ratioPresent": 0.919107000815105, "nbUnion": 35367, "duplicates": 190, "invented": 87 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 483 }, "info": { "protonNMR": { "present": 595, "notPresent": 47, "valid": 233, "notValid": 362 }, "carbonNMR": { "present": 440, "notPresent": 202, "valid": 170, "notValid": 270 }, "elementalAnalysis": { "present": 262, "notPresent": 380, "valid": 215, "notValid": 47 }, "hrms": { "present": 87, "notPresent": 555, "valid": 11, "notValid": 76 } }, "logs": { "info": 75, "warn": 201, "error": 207, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447898, "totalTokens": 967649, "nbPrompts": 642, "version": "v102", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.415547 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v102" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v103.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v103" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 3" }, "stats": { "nmr": { "proton": { "nbSpectra": 595, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 5, "not specified": 1 }, "solvent": { "CDCl3": 98, "not specified": 9, "unknown": 4, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 439, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1, "not specified": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 79, "CD3OD": 31, "null": 2, "CD3COCD3": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "not specified": 5, "MeOH": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "unknown": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2066 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 16 }, { "number": 29, "counter": 4 }, { "number": 400, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 15, "counter": 3 }, { "number": 35, "counter": 3 }, { "number": 50, "counter": 3 }, { "number": 3.83, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 8, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.87, "counter": 1 }, { "number": 5.29, "counter": 1 }, { "number": 5.3, "counter": 1 }, { "number": 11.31, "counter": 1 }, { "number": 11.32, "counter": 1 }, { "number": 64.52, "counter": 1 }, { "number": 64.56, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 159 }, { "number": 1, "counter": 93 }, { "number": 14, "counter": 49 }, { "number": 10, "counter": 42 }, { "number": 16, "counter": 40 }, { "number": 18, "counter": 38 }, { "number": 8, "counter": 36 }, { "number": 9, "counter": 36 }, { "number": 4, "counter": 34 }, { "number": 6, "counter": 32 }, { "number": 15, "counter": 32 }, { "number": 12, "counter": 31 }, { "number": 5, "counter": 30 }, { "number": 17, "counter": 29 }, { "number": 3, "counter": 28 }, { "number": 2, "counter": 25 }, { "number": 24, "counter": 22 }, { "number": 11, "counter": 20 }, { "number": 19, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 21, "counter": 19 }, { "number": 23, "counter": 18 }, { "number": 20, "counter": 17 }, { "number": 28, "counter": 15 }, { "number": 22, "counter": 14 }, { "number": 100, "counter": 11 }, { "number": 211, "counter": 11 }, { "number": 224, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": 267, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": 32, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -7, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 30, "counter": 7 }, { "number": 31, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -8, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 5 } ], "valid": 535, "invalid": 107, "ratioPresent": 0.9230773350363198, "nbUnion": 35368, "duplicates": 237, "invented": 86 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 482 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 235, "notValid": 362 }, "carbonNMR": { "present": 442, "notPresent": 200, "valid": 173, "notValid": 269 }, "elementalAnalysis": { "present": 266, "notPresent": 376, "valid": 216, "notValid": 50 }, "hrms": { "present": 78, "notPresent": 564, "valid": 12, "notValid": 66 } }, "logs": { "info": 81, "warn": 199, "error": 202, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447988, "totalTokens": 967739, "nbPrompts": 642, "version": "v103", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.415727 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v103" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v104.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v104" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 4" }, "stats": { "nmr": { "proton": { "nbSpectra": 597, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 20, "500": 5 }, "solvent": { "CDCl3": 96, "not specified": 10, "CD3OD": 31, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "Not specified": 2, "null": 1, "CDC13": 1, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "unknown": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "N/A": 1, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 443, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 28, "CDCl3": 77, "Not specified": 2, "null": 1, "CD3COCD3": 1, "not specified": 9, "pyridine-d5": 7, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "N/A": 1, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2277 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 0, "counter": 7 }, { "number": 50, "counter": 5 }, { "number": 15, "counter": 3 }, { "number": 26, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": 400, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 10, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 2.64, "counter": 1 }, { "number": 13.82, "counter": 1 }, { "number": 30.71, "counter": 1 }, { "number": 52.83, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 3.57, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 168 }, { "number": 1, "counter": 97 }, { "number": 14, "counter": 41 }, { "number": 10, "counter": 39 }, { "number": 16, "counter": 38 }, { "number": 8, "counter": 36 }, { "number": 6, "counter": 33 }, { "number": 12, "counter": 32 }, { "number": 18, "counter": 32 }, { "number": 4, "counter": 31 }, { "number": 5, "counter": 31 }, { "number": 9, "counter": 30 }, { "number": 3, "counter": 28 }, { "number": 15, "counter": 27 }, { "number": 2, "counter": 26 }, { "number": 17, "counter": 25 }, { "number": 24, "counter": 24 }, { "number": 19, "counter": 20 }, { "number": 20, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 11, "counter": 18 }, { "number": 21, "counter": 17 }, { "number": 28, "counter": 12 }, { "number": -10, "counter": 12 }, { "number": -9, "counter": 12 }, { "number": -7, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": -8, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 33, "counter": 8 }, { "number": 100, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 211, "counter": 6 }, { "number": 239, "counter": 6 }, { "number": 267, "counter": 6 }, { "number": -14, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -13, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 36, "counter": 5 } ], "valid": 537, "invalid": 105, "ratioPresent": 0.9222443730567383, "nbUnion": 35369, "duplicates": 235, "invented": 96 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 483 }, "info": { "protonNMR": { "present": 599, "notPresent": 43, "valid": 237, "notValid": 362 }, "carbonNMR": { "present": 444, "notPresent": 198, "valid": 172, "notValid": 272 }, "elementalAnalysis": { "present": 265, "notPresent": 377, "valid": 218, "notValid": 47 }, "hrms": { "present": 82, "notPresent": 560, "valid": 15, "notValid": 67 } }, "logs": { "info": 74, "warn": 200, "error": 209, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447819, "totalTokens": 967570, "nbPrompts": 642, "version": "v104", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.415389 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v104" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v105.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v105" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 598, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 5, "Not specified": 1 }, "solvent": { "CDCl3": 96, "not specified": 14, "CD3OD": 35, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 5, "null": 1, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "Not specified": 2, "MeOD": 3, "unknown": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 443, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "Not specified": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "not specified": 10, "CD3OD": 32, "CDCl3": 76, "MeOH": 4, "null": 1, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "Not specified": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1993 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 15 }, { "number": 400, "counter": 5 }, { "number": 0, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 7.01, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 7, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.88, "counter": 1 }, { "number": 5.34, "counter": 1 }, { "number": 11.84, "counter": 1 }, { "number": 64.63, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 6.7, "counter": 1 }, { "number": 6.76, "counter": 1 }, { "number": 7.09, "counter": 1 }, { "number": 29.63, "counter": 1 }, { "number": 29.86, "counter": 1 }, { "number": 56.43, "counter": 1 }, { "number": 56.52, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 424 }, { "number": -1, "counter": 161 }, { "number": 1, "counter": 88 }, { "number": 14, "counter": 50 }, { "number": 16, "counter": 44 }, { "number": 4, "counter": 39 }, { "number": 8, "counter": 38 }, { "number": 10, "counter": 38 }, { "number": 18, "counter": 36 }, { "number": 12, "counter": 35 }, { "number": 5, "counter": 34 }, { "number": 15, "counter": 34 }, { "number": 6, "counter": 31 }, { "number": 9, "counter": 31 }, { "number": 2, "counter": 29 }, { "number": 3, "counter": 29 }, { "number": 17, "counter": 25 }, { "number": 11, "counter": 23 }, { "number": 24, "counter": 22 }, { "number": 25, "counter": 22 }, { "number": 7, "counter": 21 }, { "number": 20, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 21, "counter": 18 }, { "number": 23, "counter": 18 }, { "number": 28, "counter": 15 }, { "number": 22, "counter": 14 }, { "number": 27, "counter": 12 }, { "number": 100, "counter": 12 }, { "number": 211, "counter": 10 }, { "number": 224, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 33, "counter": 8 }, { "number": 225, "counter": 8 }, { "number": -7, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 36, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": -8, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": 0, "counter": 5 } ], "valid": 525, "invalid": 117, "ratioPresent": 0.9189130192923364, "nbUnion": 35379, "duplicates": 280, "invented": 108 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 484 }, "info": { "protonNMR": { "present": 599, "notPresent": 43, "valid": 230, "notValid": 369 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 174, "notValid": 269 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 219, "notValid": 45 }, "hrms": { "present": 88, "notPresent": 554, "valid": 15, "notValid": 73 } }, "logs": { "info": 75, "warn": 205, "error": 204, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449212, "totalTokens": 968963, "nbPrompts": 642, "version": "v105", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.418175 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v105" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v106.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v106" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 3 }, "solvent": { "CDCl3": 93, "not specified": 9, "uncertain": 1, "CD3OD": 34, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 8, "null": 2, "CDC13": 1, "N/A": 1, "unknown": 2, "pyridine-d5": 7, "Not specified": 3, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "unspecified": 1, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 438, "frequency": { "50": 2, "75": 5, "80": 2, "100": 18, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 77, "CD3OD": 31, "null": 2, "CD3COCD3": 1, "N/A": 1, "unknown": 1, "pyridine-d5": 7, "not specified": 6, "MeOH": 5, "Not specified": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2189 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 20 }, { "number": 15, "counter": 5 }, { "number": 29, "counter": 4 }, { "number": 400, "counter": 3 }, { "number": -18, "counter": 2 }, { "number": 0, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 11, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 3.05, "counter": 1 }, { "number": 3.18, "counter": 1 }, { "number": 8.14, "counter": 1 }, { "number": 8.27, "counter": 1 }, { "number": 28.03, "counter": 1 }, { "number": 28.21, "counter": 1 }, { "number": 60.47, "counter": 1 }, { "number": 60.65, "counter": 1 }, { "number": 33.3, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 423 }, { "number": -1, "counter": 164 }, { "number": 1, "counter": 95 }, { "number": 14, "counter": 48 }, { "number": 8, "counter": 39 }, { "number": 10, "counter": 39 }, { "number": 4, "counter": 35 }, { "number": 16, "counter": 34 }, { "number": 9, "counter": 33 }, { "number": 12, "counter": 33 }, { "number": 15, "counter": 33 }, { "number": 2, "counter": 31 }, { "number": 18, "counter": 31 }, { "number": 5, "counter": 30 }, { "number": 6, "counter": 30 }, { "number": 3, "counter": 27 }, { "number": 17, "counter": 26 }, { "number": 24, "counter": 22 }, { "number": 25, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 20, "counter": 16 }, { "number": 100, "counter": 15 }, { "number": 21, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 22, "counter": 11 }, { "number": 26, "counter": 9 }, { "number": 27, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 211, "counter": 8 }, { "number": 31, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": 239, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": -10, "counter": 7 }, { "number": -7, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": -9, "counter": 6 }, { "number": -8, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 5 }, { "number": 32, "counter": 5 }, { "number": 254, "counter": 5 }, { "number": 0.1, "counter": 5 } ], "valid": 523, "invalid": 119, "ratioPresent": 0.9203433580429633, "nbUnion": 35384, "duplicates": 257, "invented": 106 }, "meta": { "withName": 640, "withoutName": 2, "withMolfile": 483 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 239, "notValid": 358 }, "carbonNMR": { "present": 439, "notPresent": 203, "valid": 171, "notValid": 268 }, "elementalAnalysis": { "present": 265, "notPresent": 377, "valid": 215, "notValid": 50 }, "hrms": { "present": 86, "notPresent": 556, "valid": 13, "notValid": 73 } }, "logs": { "info": 80, "warn": 192, "error": 211, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447889, "totalTokens": 967640, "nbPrompts": 642, "version": "v106", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.415529 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v106" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v107.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v107" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 7" }, "stats": { "nmr": { "proton": { "nbSpectra": 593, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 6, "not specified": 1 }, "solvent": { "CDCl3": 94, "not specified": 15, "unknown": 6, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "CDC13": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 5, "Not specified": 2, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 440, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "not specified": 12, "CDCl3": 78, "CD3OD": 31, "unknown": 4, "null": 1, "CD3COCD3": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 2, "Not specified": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 2104 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 19 }, { "number": 15, "counter": 5 }, { "number": 400, "counter": 4 }, { "number": 3.15, "counter": 3 }, { "number": 0, "counter": 2 }, { "number": 24, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 123, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 296.5, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 7.59, "counter": 1 }, { "number": 79.66, "counter": 1 }, { "number": 7.82, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 91 }, { "number": 14, "counter": 50 }, { "number": 10, "counter": 45 }, { "number": 15, "counter": 42 }, { "number": 9, "counter": 39 }, { "number": 18, "counter": 36 }, { "number": 6, "counter": 35 }, { "number": 16, "counter": 35 }, { "number": 5, "counter": 34 }, { "number": 8, "counter": 34 }, { "number": 12, "counter": 34 }, { "number": 2, "counter": 33 }, { "number": 4, "counter": 33 }, { "number": 3, "counter": 31 }, { "number": 17, "counter": 26 }, { "number": 24, "counter": 22 }, { "number": 7, "counter": 20 }, { "number": 11, "counter": 20 }, { "number": 19, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 23, "counter": 18 }, { "number": 20, "counter": 17 }, { "number": 28, "counter": 16 }, { "number": 21, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": -4, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 27, "counter": 10 }, { "number": -10, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 22, "counter": 9 }, { "number": -5, "counter": 9 }, { "number": 211, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 32, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 224, "counter": 7 }, { "number": 239, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 267, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": 0, "counter": 5 } ], "valid": 529, "invalid": 113, "ratioPresent": 0.9176275021561525, "nbUnion": 35373, "duplicates": 283, "invented": 99 }, "meta": { "withName": 637, "withoutName": 5, "withMolfile": 482 }, "info": { "protonNMR": { "present": 595, "notPresent": 47, "valid": 231, "notValid": 364 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 171, "notValid": 272 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 214, "notValid": 47 }, "hrms": { "present": 95, "notPresent": 547, "valid": 12, "notValid": 83 } }, "logs": { "info": 78, "warn": 197, "error": 207, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449269, "totalTokens": 969020, "nbPrompts": 642, "version": "v107", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.4182890000000001 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v107" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v108.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v108" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt. Replicate 8" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 20, "500": 4, "not specified": 1 }, "solvent": { "CDCl3": 91, "not specified": 16, "CD3OD": 34, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unknown": 4, "null": 1, "CDC13": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 5, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 444, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "not specified": 11, "CD3OD": 31, "CDCl3": 72, "unknown": 2, "null": 1, "CD3COCD3": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2157 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 0, "counter": 3 }, { "number": 15, "counter": 3 }, { "number": 28, "counter": 2 }, { "number": 400, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 3.38, "counter": 1 }, { "number": 8.01, "counter": 1 }, { "number": 10.76, "counter": 1 }, { "number": 77.85, "counter": 1 }, { "number": 7.5, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 7.27, "counter": 1 }, { "number": 8.42, "counter": 1 }, { "number": 11.58, "counter": 1 }, { "number": 72.73, "counter": 1 }, { "number": 6.99, "counter": 1 }, { "number": 8.83, "counter": 1 }, { "number": 12.23, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 161 }, { "number": 1, "counter": 94 }, { "number": 14, "counter": 53 }, { "number": 10, "counter": 43 }, { "number": 15, "counter": 41 }, { "number": 16, "counter": 39 }, { "number": 8, "counter": 38 }, { "number": 9, "counter": 36 }, { "number": 18, "counter": 34 }, { "number": 4, "counter": 33 }, { "number": 5, "counter": 33 }, { "number": 12, "counter": 33 }, { "number": 6, "counter": 32 }, { "number": 2, "counter": 30 }, { "number": 3, "counter": 30 }, { "number": 17, "counter": 25 }, { "number": 11, "counter": 23 }, { "number": 24, "counter": 22 }, { "number": 7, "counter": 21 }, { "number": 20, "counter": 21 }, { "number": 25, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 28, "counter": 19 }, { "number": 23, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 22, "counter": 12 }, { "number": 100, "counter": 12 }, { "number": 26, "counter": 11 }, { "number": 27, "counter": 11 }, { "number": 211, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": 267, "counter": 10 }, { "number": 239, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 282, "counter": 8 }, { "number": -9, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": -10, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": -8, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": 32, "counter": 5 }, { "number": 34, "counter": 5 }, { "number": 70, "counter": 5 }, { "number": 251, "counter": 5 }, { "number": 254, "counter": 5 } ], "valid": 537, "invalid": 105, "ratioPresent": 0.9187055630150113, "nbUnion": 35353, "duplicates": 175, "invented": 75 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 485 }, "info": { "protonNMR": { "present": 596, "notPresent": 46, "valid": 232, "notValid": 364 }, "carbonNMR": { "present": 444, "notPresent": 198, "valid": 172, "notValid": 272 }, "elementalAnalysis": { "present": 265, "notPresent": 377, "valid": 221, "notValid": 44 }, "hrms": { "present": 81, "notPresent": 561, "valid": 13, "notValid": 68 } }, "logs": { "info": 83, "warn": 200, "error": 202, "fatal": 157 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449135, "totalTokens": 968886, "nbPrompts": 642, "version": "v108", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.418021 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0.2", "version": "v108" } ], "versions": [ "v1", "v101", "v102", "v103", "v104", "v105", "v106", "v107", "v108" ], "stats": { "yaml": { "parsable": [ 639, 639, 641, 642, 642, 640, 641, 640, 641 ], "notParsable": [ 3, 3, 1, 0, 0, 2, 1, 2, 1 ], "empty": [ 2015, 2222, 2064, 2066, 2277, 1993, 2189, 2104, 2157 ] }, "numbers": { "valid": [ 524, 526, 534, 535, 537, 525, 523, 529, 537 ], "invalid": [ 118, 116, 108, 107, 105, 117, 119, 113, 105 ], "ratioPresent": [ 0.9223823368962465, 0.9152301112083121, 0.919107000815105, 0.9230773350363198, 0.9222443730567383, 0.9189130192923364, 0.9203433580429633, 0.9176275021561525, 0.9187055630150113 ], "nbUnion": [ 35374, 35369, 35367, 35368, 35369, 35379, 35384, 35373, 35353 ], "duplicates": [ 206, 248, 190, 237, 235, 280, 257, 283, 175 ], "invented": [ 99, 86, 87, 86, 96, 108, 106, 99, 75 ] }, "meta": { "withName": [ 635, 637, 638, 639, 639, 638, 640, 637, 639 ], "withoutName": [ 7, 5, 4, 3, 3, 4, 2, 5, 3 ], "withMolfile": [ 481, 481, 483, 482, 483, 484, 483, 482, 485 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.2_0.2", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 570, "frequency": { "50": 4, "80": 1, "200": 3, "300": 11, "400": 45, "500": 7 }, "solvent": { "unknown": 88, "not specified": 13, "ethanol-water": 1, "CDCl3": 97, "CD3OD": 30, "CHCl3": 2, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "N/A": 4, "pyridine-d5": 7, "MeOH": 4, "methanol": 1, "ppm": 2, "MeOD": 3, "dimethylformamide": 1, "dioxane": 1, "δ/ppm": 1, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 482, "frequency": { "50": 2, "75": 5, "80": 1, "100": 45, "125": 3, "200": 1, "300": 1, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "unknown": 74, "DMSO-d6": 55, "not specified": 12, "CD3OD": 28, "CHCl3": 2, "CDCl3": 88, "CD3COCD3": 1, "N/A": 4, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "MeOH": 2, "ppm": 1 } } }, "yaml": { "parsable": 629, "notParsable": 13, "empty": 15781 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 156 }, { "number": 400, "counter": 28 }, { "number": 100, "counter": 11 }, { "number": -18, "counter": 2 }, { "number": 17, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 439 }, { "number": -1, "counter": 140 }, { "number": 1, "counter": 73 }, { "number": 14, "counter": 51 }, { "number": 100, "counter": 43 }, { "number": 4, "counter": 42 }, { "number": 18, "counter": 42 }, { "number": 5, "counter": 41 }, { "number": 8, "counter": 38 }, { "number": 17, "counter": 37 }, { "number": 15, "counter": 36 }, { "number": 3, "counter": 33 }, { "number": 10, "counter": 33 }, { "number": 16, "counter": 33 }, { "number": 2, "counter": 30 }, { "number": 12, "counter": 30 }, { "number": 6, "counter": 29 }, { "number": 11, "counter": 28 }, { "number": 9, "counter": 27 }, { "number": 24, "counter": 26 }, { "number": 19, "counter": 24 }, { "number": 25, "counter": 24 }, { "number": 21, "counter": 19 }, { "number": 7, "counter": 18 }, { "number": 20, "counter": 17 }, { "number": 224, "counter": 17 }, { "number": 28, "counter": 16 }, { "number": 211, "counter": 16 }, { "number": 239, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": 23, "counter": 15 }, { "number": 22, "counter": 14 }, { "number": 196, "counter": 14 }, { "number": 27, "counter": 13 }, { "number": 30, "counter": 13 }, { "number": 43, "counter": 12 }, { "number": 26, "counter": 11 }, { "number": 282, "counter": 11 }, { "number": 32, "counter": 10 }, { "number": 70, "counter": 10 }, { "number": 254, "counter": 10 }, { "number": 31, "counter": 9 }, { "number": 251, "counter": 9 }, { "number": 33, "counter": 8 }, { "number": 135, "counter": 8 }, { "number": 237, "counter": 8 }, { "number": -13, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": -3, "counter": 8 }, { "number": 45, "counter": 7 } ], "valid": 383, "invalid": 259, "ratioPresent": 0.899111641765576, "nbUnion": 35505, "duplicates": 272, "invented": 636 }, "meta": { "withName": 622, "withoutName": 20, "withMolfile": 474 }, "info": { "protonNMR": { "present": 587, "notPresent": 55, "valid": 166, "notValid": 421 }, "carbonNMR": { "present": 518, "notPresent": 124, "valid": 150, "notValid": 368 }, "elementalAnalysis": { "present": 279, "notPresent": 363, "valid": 232, "notValid": 47 }, "hrms": { "present": 58, "notPresent": 584, "valid": 8, "notValid": 50 } }, "logs": { "info": 37, "warn": 192, "error": 245, "fatal": 168 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 532701, "totalTokens": 1019710, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.552411 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.2_0.2", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 629 ], "notParsable": [ 13 ], "empty": [ 15781 ] }, "numbers": { "valid": [ 383 ], "invalid": [ 259 ], "ratioPresent": [ 0.899111641765576 ], "nbUnion": [ 35505 ], "duplicates": [ 272 ], "invented": [ 636 ] }, "meta": { "withName": [ 622 ], "withoutName": [ 20 ], "withMolfile": [ 474 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.2_0.2", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 581, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 31, "500": 4, "6.8": 1, "400\"": 1, "7.57": 1 }, "solvent": { "CDCl3": 84, "not specified": 7, "N/A": 7, "DMSO-d6": 73, "unknown": 13, "CD3OD": 31, "δ/ppm": 1, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "literature": 1, "pyridine-d5": 6, "MeOH": 9, "MeOD": 3, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 3, "acetone-d6": 3, "δ": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 430, "frequency": { "50": 2, "75": 5, "80": 2, "100": 26, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1, "100\"": 1, "166.8": 1 }, "solvent": { "DMSO-d6": 52, "unknown": 10, "CD3OD": 29, "CDCl3": 69, "CHCl3": 2, "CD3COCD3": 1, "N/A": 5, "literature": 1, "pyridine-d5": 6, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "not specified": 5, "ppm": 1, "D2O": 1, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "δ": 1, "CH3OH": 1 } } }, "yaml": { "parsable": 626, "notParsable": 16, "empty": 2959 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 14 }, { "number": 100, "counter": 14 }, { "number": 400, "counter": 14 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 13, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 435 }, { "number": -1, "counter": 172 }, { "number": 1, "counter": 78 }, { "number": 14, "counter": 59 }, { "number": 3, "counter": 47 }, { "number": 16, "counter": 47 }, { "number": 10, "counter": 43 }, { "number": 4, "counter": 42 }, { "number": 15, "counter": 42 }, { "number": 8, "counter": 41 }, { "number": 18, "counter": 41 }, { "number": 2, "counter": 37 }, { "number": 6, "counter": 37 }, { "number": 5, "counter": 36 }, { "number": 12, "counter": 34 }, { "number": 9, "counter": 32 }, { "number": 17, "counter": 30 }, { "number": 24, "counter": 28 }, { "number": 19, "counter": 25 }, { "number": 25, "counter": 25 }, { "number": 7, "counter": 23 }, { "number": 21, "counter": 22 }, { "number": -10, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 23, "counter": 21 }, { "number": 20, "counter": 19 }, { "number": 22, "counter": 18 }, { "number": -7, "counter": 18 }, { "number": -8, "counter": 16 }, { "number": 100, "counter": 14 }, { "number": -13, "counter": 13 }, { "number": -9, "counter": 13 }, { "number": 28, "counter": 12 }, { "number": 26, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 33, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": -2, "counter": 9 }, { "number": -3, "counter": 8 }, { "number": -12, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 0, "counter": 7 }, { "number": 30, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": -11, "counter": 6 }, { "number": 36, "counter": 5 }, { "number": 43, "counter": 5 }, { "number": 55, "counter": 5 }, { "number": 70, "counter": 5 }, { "number": 209, "counter": 5 } ], "valid": 505, "invalid": 137, "ratioPresent": 0.9064777732799018, "nbUnion": 35352, "duplicates": 320, "invented": 94 }, "meta": { "withName": 610, "withoutName": 32, "withMolfile": 457 }, "info": { "protonNMR": { "present": 581, "notPresent": 61, "valid": 227, "notValid": 354 }, "carbonNMR": { "present": 430, "notPresent": 212, "valid": 162, "notValid": 268 }, "elementalAnalysis": { "present": 251, "notPresent": 391, "valid": 208, "notValid": 43 }, "hrms": { "present": 82, "notPresent": 560, "valid": 13, "notValid": 69 } }, "logs": { "info": 66, "warn": 207, "error": 184, "fatal": 185 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 470587, "totalTokens": 947324, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.417911 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.2_0.2", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 626 ], "notParsable": [ 16 ], "empty": [ 2959 ] }, "numbers": { "valid": [ 505 ], "invalid": [ 137 ], "ratioPresent": [ 0.9064777732799018 ], "nbUnion": [ 35352 ], "duplicates": [ 320 ], "invented": [ 94 ] }, "meta": { "withName": [ 610 ], "withoutName": [ 32 ], "withMolfile": [ 457 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_0.2_0.2", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 555, "frequency": { "20": 1, "50": 4, "80": 2, "200": 4, "300": 8, "400": 41, "500": 3, "Not specified": 1 }, "solvent": { "CDCl3": 89, "DMSO-d6": 67, "not specified": 39, "CD3OD": 33, "Not specified": 11, "unknown": 3, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 5, "MeOH": 7, "CDC13": 1, "pyridine-d5": 7, "MeOD": 3, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "δ": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 441, "frequency": { "20": 1, "50": 2, "75": 3, "80": 2, "100": 36, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1, "174.9": 1 }, "solvent": { "DMSO-d6": 51, "not specified": 29, "DEPT 135": 1, "CD3OD": 31, "CDCl3": 74, "unknown": 2, "Not specified": 8, "CHCl3": 3, "MeOH": 7, "CD3COCD3": 1, "pyridine-d5": 7, "CD3CN": 3, "MeOD": 3, "DMSO": 1, "ppm": 1, "D2O": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "δ": 1, "CH3OH": 1 } } }, "yaml": { "parsable": 593, "notParsable": 49, "empty": 2446 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 31 }, { "number": 400, "counter": 24 }, { "number": 0, "counter": 13 }, { "number": 50, "counter": 6 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 2, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 11, "counter": 1 }, { "number": 13, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 16, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 4.6, "counter": 1 }, { "number": 4.62, "counter": 1 }, { "number": 7.68, "counter": 1 }, { "number": 7.72, "counter": 1 }, { "number": 26.74, "counter": 1 }, { "number": 26.81, "counter": 1 }, { "number": 60.85, "counter": 1 }, { "number": 60.98, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": -3, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 193 }, { "number": 1, "counter": 131 }, { "number": 3, "counter": 77 }, { "number": 2, "counter": 71 }, { "number": 4, "counter": 67 }, { "number": 6, "counter": 60 }, { "number": 5, "counter": 59 }, { "number": 8, "counter": 53 }, { "number": 100, "counter": 47 }, { "number": 10, "counter": 44 }, { "number": 7, "counter": 41 }, { "number": 14, "counter": 39 }, { "number": 15, "counter": 33 }, { "number": 12, "counter": 32 }, { "number": -3, "counter": 31 }, { "number": 11, "counter": 26 }, { "number": 16, "counter": 26 }, { "number": 18, "counter": 26 }, { "number": 25, "counter": 26 }, { "number": 9, "counter": 25 }, { "number": 17, "counter": 24 }, { "number": 23, "counter": 23 }, { "number": 24, "counter": 21 }, { "number": -2, "counter": 21 }, { "number": -4, "counter": 21 }, { "number": 19, "counter": 17 }, { "number": 50, "counter": 15 }, { "number": 21, "counter": 14 }, { "number": 90, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 63, "counter": 12 }, { "number": 224, "counter": 12 }, { "number": 27, "counter": 11 }, { "number": -5, "counter": 11 }, { "number": 20, "counter": 10 }, { "number": 22, "counter": 10 }, { "number": 237, "counter": 10 }, { "number": 360, "counter": 10 }, { "number": 70, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 91, "counter": 8 }, { "number": 105, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 225, "counter": 8 }, { "number": 310, "counter": 8 }, { "number": 0, "counter": 7 }, { "number": 34, "counter": 7 }, { "number": 116, "counter": 7 }, { "number": 239, "counter": 7 } ], "valid": 498, "invalid": 144, "ratioPresent": 0.8625267225321234, "nbUnion": 35401, "duplicates": 257, "invented": 140 }, "meta": { "withName": 582, "withoutName": 60, "withMolfile": 441 }, "info": { "protonNMR": { "present": 556, "notPresent": 86, "valid": 195, "notValid": 361 }, "carbonNMR": { "present": 441, "notPresent": 201, "valid": 158, "notValid": 283 }, "elementalAnalysis": { "present": 237, "notPresent": 405, "valid": 196, "notValid": 41 }, "hrms": { "present": 96, "notPresent": 546, "valid": 9, "notValid": 87 } }, "logs": { "info": 65, "warn": 146, "error": 230, "fatal": 201 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 470349, "totalTokens": 944518, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.414867 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_0.2_0.2", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 593 ], "notParsable": [ 49 ], "empty": [ 2446 ] }, "numbers": { "valid": [ 498 ], "invalid": [ 144 ], "ratioPresent": [ 0.8625267225321234 ], "nbUnion": [ 35401 ], "duplicates": [ 257 ], "invented": [ 140 ] }, "meta": { "withName": [ 582 ], "withoutName": [ 60 ], "withMolfile": [ 441 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_0.2_0.2", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0.2_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0.2_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 588, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 3 }, "solvent": { "CDCl3": 88, "Not specified": 1, "DMSO-d6": 74, "CD3OD": 32, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 7, "CDC13": 1, "pyridine-d5": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "water/methanol": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 435, "frequency": { "50": 2, "75": 5, "80": 2, "100": 18, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 29, "CHCl3": 1, "CDCl3": 72, "CD3COCD3": 1, "pyridine-d5": 6, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "CH3OH": 1 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1807 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 13 }, { "number": 0, "counter": 11 }, { "number": 400, "counter": 3 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 15, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 7.5, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 436 }, { "number": -1, "counter": 179 }, { "number": 1, "counter": 74 }, { "number": 14, "counter": 51 }, { "number": 16, "counter": 45 }, { "number": 4, "counter": 41 }, { "number": 10, "counter": 41 }, { "number": 15, "counter": 38 }, { "number": 8, "counter": 37 }, { "number": 9, "counter": 37 }, { "number": 3, "counter": 36 }, { "number": 5, "counter": 34 }, { "number": 18, "counter": 32 }, { "number": 6, "counter": 30 }, { "number": 12, "counter": 29 }, { "number": 17, "counter": 28 }, { "number": 2, "counter": 26 }, { "number": 11, "counter": 25 }, { "number": 7, "counter": 24 }, { "number": 24, "counter": 24 }, { "number": 23, "counter": 23 }, { "number": 25, "counter": 23 }, { "number": 19, "counter": 21 }, { "number": -10, "counter": 19 }, { "number": 20, "counter": 17 }, { "number": 22, "counter": 17 }, { "number": 21, "counter": 16 }, { "number": -7, "counter": 16 }, { "number": 28, "counter": 15 }, { "number": 100, "counter": 14 }, { "number": -9, "counter": 14 }, { "number": -8, "counter": 14 }, { "number": 30, "counter": 10 }, { "number": -2, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": -3, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": -13, "counter": 8 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 29, "counter": 6 }, { "number": 34, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 211, "counter": 6 }, { "number": 224, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 0.1, "counter": 6 }, { "number": 4.2, "counter": 6 }, { "number": -6, "counter": 6 } ], "valid": 529, "invalid": 113, "ratioPresent": 0.9133546015739121, "nbUnion": 35334, "duplicates": 263, "invented": 67 }, "meta": { "withName": 617, "withoutName": 25, "withMolfile": 462 }, "info": { "protonNMR": { "present": 588, "notPresent": 54, "valid": 230, "notValid": 358 }, "carbonNMR": { "present": 435, "notPresent": 207, "valid": 171, "notValid": 264 }, "elementalAnalysis": { "present": 248, "notPresent": 394, "valid": 211, "notValid": 37 }, "hrms": { "present": 80, "notPresent": 562, "valid": 13, "notValid": 67 } }, "logs": { "info": 68, "warn": 214, "error": 180, "fatal": 180 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 439350, "totalTokens": 921865, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0.2, "cost": 1.361215 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_0.2_0.2", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 640 ], "notParsable": [ 2 ], "empty": [ 1807 ] }, "numbers": { "valid": [ 529 ], "invalid": [ 113 ], "ratioPresent": [ 0.9133546015739121 ], "nbUnion": [ 35334 ], "duplicates": [ 263 ], "invented": [ 67 ] }, "meta": { "withName": [ 617 ], "withoutName": [ 25 ], "withMolfile": [ 462 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 593, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 27, "500": 4, "not specified": 1 }, "solvent": { "CDCl3": 99, "not specified": 13, "uncertain": 1, "CD3OD": 32, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "unknown": 3, "null": 1, "CDC13": 1, "N/A": 3, "literature": 1, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "Not specified": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 435, "frequency": { "50": 2, "75": 5, "80": 2, "100": 23, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 48, "uncertain": 1, "CD3OD": 29, "CDCl3": 79, "unknown": 2, "null": 1, "CD3COCD3": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 6, "not specified": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "Not specified": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "CH3OH": 1 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1986 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 16 }, { "number": 400, "counter": 9 }, { "number": 0, "counter": 3 }, { "number": 30, "counter": 3 }, { "number": 50, "counter": 3 }, { "number": 41, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 8, "counter": 1 }, { "number": 11, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 485, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 5.05, "counter": 1 }, { "number": 5.18, "counter": 1 }, { "number": 16.21, "counter": 1 }, { "number": 16.33, "counter": 1 }, { "number": 17.49, "counter": 1 }, { "number": 17.64, "counter": 1 }, { "number": 60.98, "counter": 1 }, { "number": 61.12, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 4.77, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 436 }, { "number": -1, "counter": 160 }, { "number": 1, "counter": 93 }, { "number": 14, "counter": 49 }, { "number": 15, "counter": 42 }, { "number": 10, "counter": 41 }, { "number": 16, "counter": 41 }, { "number": 8, "counter": 37 }, { "number": 18, "counter": 36 }, { "number": 4, "counter": 35 }, { "number": 12, "counter": 34 }, { "number": 5, "counter": 31 }, { "number": 6, "counter": 31 }, { "number": 17, "counter": 31 }, { "number": 3, "counter": 30 }, { "number": 9, "counter": 30 }, { "number": 24, "counter": 26 }, { "number": 2, "counter": 24 }, { "number": 7, "counter": 23 }, { "number": 20, "counter": 22 }, { "number": 19, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 21, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 28, "counter": 14 }, { "number": 100, "counter": 12 }, { "number": 22, "counter": 11 }, { "number": 211, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": -10, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": -8, "counter": 9 }, { "number": 32, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -7, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 27, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 30, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 254, "counter": 6 } ], "valid": 529, "invalid": 113, "ratioPresent": 0.9157177263296138, "nbUnion": 35384, "duplicates": 199, "invented": 103 }, "meta": { "withName": 637, "withoutName": 5, "withMolfile": 481 }, "info": { "protonNMR": { "present": 594, "notPresent": 48, "valid": 230, "notValid": 364 }, "carbonNMR": { "present": 435, "notPresent": 207, "valid": 168, "notValid": 267 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 217, "notValid": 47 }, "hrms": { "present": 86, "notPresent": 556, "valid": 13, "notValid": 73 } }, "logs": { "info": 72, "warn": 199, "error": 210, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447474, "totalTokens": 967225, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.414699 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v1" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v101.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v101" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 1" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 6, "Not specified": 1 }, "solvent": { "CDCl3": 96, "not specified": 11, "DMSO-d6": 74, "uncertain": 1, "CD3OD": 33, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 7, "null": 1, "unknown": 2, "CDC13": 1, "N/A": 2, "Not specified": 5, "pyridine-d5": 7, "MeOD": 3, "unspecified": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 447, "frequency": { "50": 2, "75": 5, "80": 2, "100": 23, "125": 3, "200": 2, "Not specified": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "uncertain": 1, "CD3OD": 31, "CDCl3": 79, "MeOH": 7, "null": 1, "unknown": 1, "CD3COCD3": 1, "N/A": 2, "Not specified": 5, "pyridine-d5": 7, "not specified": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1826 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 14 }, { "number": 400, "counter": 7 }, { "number": 0, "counter": 5 }, { "number": 15, "counter": 5 }, { "number": 28, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 32, "counter": 2 }, { "number": 37, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1600, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3400, "counter": 1 }, { "number": 5.05, "counter": 1 }, { "number": 5.15, "counter": 1 }, { "number": 16.39, "counter": 1 }, { "number": 16.4, "counter": 1 }, { "number": 17.68, "counter": 1 }, { "number": 17.81, "counter": 1 }, { "number": 60.65, "counter": 1 }, { "number": 60.87, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 161 }, { "number": 1, "counter": 88 }, { "number": 14, "counter": 47 }, { "number": 10, "counter": 44 }, { "number": 15, "counter": 37 }, { "number": 4, "counter": 36 }, { "number": 8, "counter": 36 }, { "number": 16, "counter": 36 }, { "number": 18, "counter": 36 }, { "number": 6, "counter": 34 }, { "number": 9, "counter": 33 }, { "number": 12, "counter": 33 }, { "number": 5, "counter": 32 }, { "number": 2, "counter": 31 }, { "number": 17, "counter": 30 }, { "number": 3, "counter": 29 }, { "number": 11, "counter": 24 }, { "number": 24, "counter": 23 }, { "number": 25, "counter": 23 }, { "number": 7, "counter": 21 }, { "number": 21, "counter": 18 }, { "number": 23, "counter": 17 }, { "number": 19, "counter": 16 }, { "number": 20, "counter": 16 }, { "number": 22, "counter": 15 }, { "number": 100, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 224, "counter": 12 }, { "number": 211, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": -10, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 30, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": -5, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": 33, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -2, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 34, "counter": 6 }, { "number": 282, "counter": 6 } ], "valid": 527, "invalid": 115, "ratioPresent": 0.9194799843169938, "nbUnion": 35378, "duplicates": 234, "invented": 103 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 484 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 227, "notValid": 370 }, "carbonNMR": { "present": 447, "notPresent": 195, "valid": 173, "notValid": 274 }, "elementalAnalysis": { "present": 266, "notPresent": 376, "valid": 218, "notValid": 48 }, "hrms": { "present": 93, "notPresent": 549, "valid": 12, "notValid": 81 } }, "logs": { "info": 78, "warn": 200, "error": 206, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448451, "totalTokens": 968202, "nbPrompts": 642, "version": "v101", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.4166530000000002 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v101" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v102.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v102" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 2" }, "stats": { "nmr": { "proton": { "nbSpectra": 597, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 25, "500": 4, "frequency": 1 }, "solvent": { "CDCl3": 99, "uncertain": 2, "not specified": 14, "CD3OD": 33, "unknown": 4, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 8, "null": 2, "CDC13": 1, "N/A": 1, "Not specified": 5, "pyridine-d5": 7, "unspecified": 1, "MeOD": 3, "Not known": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "[solvent]": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 444, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "not specified": 9, "uncertain": 1, "CD3OD": 30, "unknown": 3, "CDCl3": 79, "null": 1, "CD3COCD3": 1, "N/A": 1, "Not specified": 5, "pyridine-d5": 7, "MeOH": 3, "unspecified": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "Not known": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 1982 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 12 }, { "number": 400, "counter": 7 }, { "number": 15, "counter": 5 }, { "number": 50, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 29, "counter": 3 }, { "number": 22, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 41, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 77.65, "counter": 2 }, { "number": 7.67, "counter": 2 }, { "number": 8, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 301.1, "counter": 1 }, { "number": 3.77, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 432 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 85 }, { "number": 14, "counter": 52 }, { "number": 16, "counter": 43 }, { "number": 10, "counter": 42 }, { "number": 15, "counter": 39 }, { "number": 8, "counter": 36 }, { "number": 18, "counter": 36 }, { "number": 4, "counter": 34 }, { "number": 9, "counter": 34 }, { "number": 12, "counter": 33 }, { "number": 3, "counter": 31 }, { "number": 17, "counter": 31 }, { "number": 5, "counter": 30 }, { "number": 6, "counter": 30 }, { "number": 2, "counter": 26 }, { "number": 25, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 24, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 11, "counter": 17 }, { "number": 20, "counter": 15 }, { "number": 21, "counter": 15 }, { "number": 28, "counter": 13 }, { "number": 100, "counter": 11 }, { "number": -10, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 70, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 30, "counter": 7 }, { "number": 239, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 27, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 34, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 32, "counter": 5 }, { "number": 43, "counter": 5 }, { "number": 240, "counter": 5 } ], "valid": 542, "invalid": 100, "ratioPresent": 0.9245445611070909, "nbUnion": 35390, "duplicates": 253, "invented": 109 }, "meta": { "withName": 640, "withoutName": 2, "withMolfile": 483 }, "info": { "protonNMR": { "present": 598, "notPresent": 44, "valid": 226, "notValid": 372 }, "carbonNMR": { "present": 445, "notPresent": 197, "valid": 174, "notValid": 271 }, "elementalAnalysis": { "present": 268, "notPresent": 374, "valid": 218, "notValid": 50 }, "hrms": { "present": 81, "notPresent": 561, "valid": 14, "notValid": 67 } }, "logs": { "info": 80, "warn": 201, "error": 202, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 451890, "totalTokens": 971641, "nbPrompts": 642, "version": "v102", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.423531 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v102" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v103.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v103" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 3" }, "stats": { "nmr": { "proton": { "nbSpectra": 597, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 22, "500": 6 }, "solvent": { "CDCl3": 97, "not specified": 11, "CD3OD": 34, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 3, "unknown": 2, "CDC13": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 445, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "50.3": 1, "75.4": 1, "100.37": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 80, "CD3OD": 31, "null": 1, "unknown": 2, "CD3COCD3": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 5, "not specified": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2416 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 14 }, { "number": 0, "counter": 4 }, { "number": 15, "counter": 4 }, { "number": 400, "counter": 4 }, { "number": -18, "counter": 2 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 296.5, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 4.06, "counter": 1 }, { "number": 23.21, "counter": 1 }, { "number": 72.73, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": -100, "counter": 1 }, { "number": -98, "counter": 1 }, { "number": 3.15, "counter": 1 }, { "number": 7.63, "counter": 1 }, { "number": 79.96, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 98 }, { "number": 14, "counter": 48 }, { "number": 8, "counter": 45 }, { "number": 10, "counter": 41 }, { "number": 12, "counter": 38 }, { "number": 15, "counter": 37 }, { "number": 16, "counter": 37 }, { "number": 6, "counter": 36 }, { "number": 18, "counter": 35 }, { "number": 2, "counter": 34 }, { "number": 9, "counter": 34 }, { "number": 4, "counter": 31 }, { "number": 5, "counter": 30 }, { "number": 3, "counter": 24 }, { "number": 11, "counter": 24 }, { "number": 17, "counter": 22 }, { "number": 24, "counter": 22 }, { "number": 21, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 25, "counter": 18 }, { "number": 20, "counter": 17 }, { "number": 28, "counter": 15 }, { "number": 224, "counter": 13 }, { "number": 22, "counter": 12 }, { "number": 211, "counter": 12 }, { "number": 239, "counter": 12 }, { "number": 26, "counter": 11 }, { "number": -10, "counter": 11 }, { "number": 196, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 100, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": -13, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": 225, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": 33, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": -9, "counter": 6 }, { "number": -12, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 31, "counter": 5 }, { "number": 32, "counter": 5 } ], "valid": 539, "invalid": 103, "ratioPresent": 0.9219121270865928, "nbUnion": 35365, "duplicates": 194, "invented": 88 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 481 }, "info": { "protonNMR": { "present": 599, "notPresent": 43, "valid": 228, "notValid": 371 }, "carbonNMR": { "present": 446, "notPresent": 196, "valid": 169, "notValid": 277 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 216, "notValid": 48 }, "hrms": { "present": 84, "notPresent": 558, "valid": 14, "notValid": 70 } }, "logs": { "info": 79, "warn": 196, "error": 206, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450526, "totalTokens": 970277, "nbPrompts": 642, "version": "v103", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.420803 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v103" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v104.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v104" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 4" }, "stats": { "nmr": { "proton": { "nbSpectra": 592, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 6 }, "solvent": { "CDCl3": 96, "not specified": 10, "CD3OD": 31, "uncertain": 1, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 3, "CDC13": 1, "N/A": 1, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "unknown": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 434, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "50.3": 1, "75.4": 1, "150.9": 1 }, "solvent": { "DMSO-d6": 50, "CDCl3": 77, "CD3OD": 29, "uncertain": 1, "null": 2, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 7, "MeOH": 5, "not specified": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2208 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 19 }, { "number": 0, "counter": 6 }, { "number": 400, "counter": 5 }, { "number": 50, "counter": 4 }, { "number": 15, "counter": 2 }, { "number": 21, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 150.9, "counter": 1 }, { "number": 3.16, "counter": 1 }, { "number": 9.18, "counter": 1 }, { "number": 79.66, "counter": 1 }, { "number": 3.14, "counter": 1 }, { "number": 8.72, "counter": 1 }, { "number": 10.83, "counter": 1 }, { "number": 77.31, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 430 }, { "number": -1, "counter": 167 }, { "number": 1, "counter": 91 }, { "number": 14, "counter": 48 }, { "number": 10, "counter": 45 }, { "number": 16, "counter": 43 }, { "number": 8, "counter": 40 }, { "number": 15, "counter": 38 }, { "number": 9, "counter": 37 }, { "number": 3, "counter": 34 }, { "number": 18, "counter": 34 }, { "number": 4, "counter": 33 }, { "number": 6, "counter": 33 }, { "number": 12, "counter": 32 }, { "number": 2, "counter": 31 }, { "number": 5, "counter": 30 }, { "number": 17, "counter": 29 }, { "number": 11, "counter": 26 }, { "number": 23, "counter": 22 }, { "number": 24, "counter": 22 }, { "number": 25, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 20, "counter": 17 }, { "number": 21, "counter": 17 }, { "number": 22, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": -10, "counter": 12 }, { "number": 27, "counter": 11 }, { "number": -9, "counter": 11 }, { "number": 100, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": 239, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -5, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 70, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 5 }, { "number": 30, "counter": 5 } ], "valid": 527, "invalid": 115, "ratioPresent": 0.9160045890062518, "nbUnion": 35383, "duplicates": 215, "invented": 107 }, "meta": { "withName": 636, "withoutName": 6, "withMolfile": 481 }, "info": { "protonNMR": { "present": 593, "notPresent": 49, "valid": 238, "notValid": 355 }, "carbonNMR": { "present": 436, "notPresent": 206, "valid": 169, "notValid": 267 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 215, "notValid": 46 }, "hrms": { "present": 79, "notPresent": 563, "valid": 13, "notValid": 66 } }, "logs": { "info": 80, "warn": 196, "error": 205, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447374, "totalTokens": 967125, "nbPrompts": 642, "version": "v104", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.4144990000000002 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v104" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v105.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v105" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 600, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 20, "500": 5 }, "solvent": { "CDCl3": 96, "not specified": 8, "unknown": 4, "CD3OD": 34, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "CDC13": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 435, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "not specified": 6, "CDCl3": 77, "CD3OD": 31, "null": 1, "CD3COCD3": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "unknown": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1874 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 20 }, { "number": 15, "counter": 4 }, { "number": 8.4, "counter": 3 }, { "number": 0, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": 400, "counter": 2 }, { "number": 3.83, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 60, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.86, "counter": 1 }, { "number": 10.73, "counter": 1 }, { "number": 11.29, "counter": 1 }, { "number": 20.73, "counter": 1 }, { "number": 21.99, "counter": 1 }, { "number": 64.15, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 435 }, { "number": -1, "counter": 162 }, { "number": 1, "counter": 96 }, { "number": 14, "counter": 49 }, { "number": 10, "counter": 43 }, { "number": 16, "counter": 43 }, { "number": 8, "counter": 38 }, { "number": 6, "counter": 37 }, { "number": 15, "counter": 37 }, { "number": 18, "counter": 37 }, { "number": 9, "counter": 34 }, { "number": 12, "counter": 33 }, { "number": 5, "counter": 32 }, { "number": 2, "counter": 31 }, { "number": 4, "counter": 28 }, { "number": 17, "counter": 27 }, { "number": 3, "counter": 26 }, { "number": 24, "counter": 26 }, { "number": 7, "counter": 24 }, { "number": 20, "counter": 23 }, { "number": 23, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 21, "counter": 18 }, { "number": 19, "counter": 16 }, { "number": 22, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": -7, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": -8, "counter": 10 }, { "number": -6, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": 70, "counter": 8 }, { "number": 211, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -5, "counter": 8 }, { "number": 239, "counter": 7 }, { "number": -10, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": 48, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": -2, "counter": 6 } ], "valid": 530, "invalid": 112, "ratioPresent": 0.9178019253763291, "nbUnion": 35398, "duplicates": 224, "invented": 121 }, "meta": { "withName": 636, "withoutName": 6, "withMolfile": 480 }, "info": { "protonNMR": { "present": 601, "notPresent": 41, "valid": 237, "notValid": 364 }, "carbonNMR": { "present": 436, "notPresent": 206, "valid": 171, "notValid": 265 }, "elementalAnalysis": { "present": 259, "notPresent": 383, "valid": 212, "notValid": 47 }, "hrms": { "present": 85, "notPresent": 557, "valid": 14, "notValid": 71 } }, "logs": { "info": 78, "warn": 196, "error": 206, "fatal": 162 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448741, "totalTokens": 968492, "nbPrompts": 642, "version": "v105", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.417233 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v105" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v106.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v106" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 5, "Not specified": 1 }, "solvent": { "CDCl3": 93, "not specified": 10, "DMSO-d6": 74, "CD3OD": 34, "CD3COCD3": 2, "CD3CN": 6, "unspecified": 2, "CDC13": 1, "N/A": 1, "unknown": 1, "pyridine-d5": 7, "Not specified": 3, "MeOH": 4, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 435, "frequency": { "50": 2, "75": 5, "80": 2, "100": 18, "125": 3, "200": 2, "100.6": 1, "Not specified": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "CDCl3": 76, "CD3OD": 31, "unspecified": 2, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 7, "Not specified": 3, "not specified": 9, "MeOH": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2133 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 4 }, { "number": 50, "counter": 4 }, { "number": 28, "counter": 3 }, { "number": 400, "counter": 3 }, { "number": 15, "counter": 2 }, { "number": 21, "counter": 2 }, { "number": 24, "counter": 2 }, { "number": 32, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": 37, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 3.15, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 10, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 60, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 100.6, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 426 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 97 }, { "number": 14, "counter": 47 }, { "number": 10, "counter": 40 }, { "number": 8, "counter": 39 }, { "number": 16, "counter": 39 }, { "number": 6, "counter": 38 }, { "number": 15, "counter": 38 }, { "number": 12, "counter": 35 }, { "number": 4, "counter": 34 }, { "number": 9, "counter": 33 }, { "number": 2, "counter": 31 }, { "number": 3, "counter": 30 }, { "number": 18, "counter": 29 }, { "number": 5, "counter": 27 }, { "number": 17, "counter": 24 }, { "number": 24, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 23, "counter": 21 }, { "number": 7, "counter": 18 }, { "number": 20, "counter": 18 }, { "number": 25, "counter": 18 }, { "number": 21, "counter": 17 }, { "number": 19, "counter": 14 }, { "number": 28, "counter": 14 }, { "number": 100, "counter": 11 }, { "number": -8, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": 27, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": -10, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": 239, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 36, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 267, "counter": 6 }, { "number": 282, "counter": 6 } ], "valid": 533, "invalid": 109, "ratioPresent": 0.9176873790725858, "nbUnion": 35393, "duplicates": 206, "invented": 121 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 483 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 233, "notValid": 364 }, "carbonNMR": { "present": 436, "notPresent": 206, "valid": 169, "notValid": 267 }, "elementalAnalysis": { "present": 267, "notPresent": 375, "valid": 219, "notValid": 48 }, "hrms": { "present": 87, "notPresent": 555, "valid": 13, "notValid": 74 } }, "logs": { "info": 74, "warn": 202, "error": 207, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448313, "totalTokens": 968064, "nbPrompts": 642, "version": "v106", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.416377 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v106" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v107.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v107" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 7" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 6 }, "solvent": { "CDCl3": 97, "not specified": 6, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 5, "Not specified": 2, "MeOD": 3, "unknown": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 439, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "CD3OD": 30, "CDCl3": 79, "null": 2, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "not specified": 3, "MeOH": 5, "Not specified": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "unknown": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2221 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 19 }, { "number": 400, "counter": 5 }, { "number": 15, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 22, "counter": 2 }, { "number": 28, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 5, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 2.49, "counter": 1 }, { "number": 2.51, "counter": 1 }, { "number": 6.61, "counter": 1 }, { "number": 6.66, "counter": 1 }, { "number": 79.68, "counter": 1 }, { "number": 79.72, "counter": 1 }, { "number": 7.91, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 164 }, { "number": 1, "counter": 91 }, { "number": 14, "counter": 46 }, { "number": 8, "counter": 45 }, { "number": 10, "counter": 43 }, { "number": 16, "counter": 40 }, { "number": 12, "counter": 35 }, { "number": 15, "counter": 34 }, { "number": 4, "counter": 33 }, { "number": 3, "counter": 32 }, { "number": 5, "counter": 32 }, { "number": 9, "counter": 32 }, { "number": 2, "counter": 31 }, { "number": 6, "counter": 31 }, { "number": 17, "counter": 28 }, { "number": 18, "counter": 27 }, { "number": 24, "counter": 26 }, { "number": 19, "counter": 24 }, { "number": 7, "counter": 21 }, { "number": 11, "counter": 21 }, { "number": 23, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 20, "counter": 16 }, { "number": 28, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 100, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": -10, "counter": 13 }, { "number": -9, "counter": 13 }, { "number": -7, "counter": 12 }, { "number": -4, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": -8, "counter": 9 }, { "number": 224, "counter": 8 }, { "number": -2, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 211, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": 239, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 7.3, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 31, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 196, "counter": 6 } ], "valid": 530, "invalid": 112, "ratioPresent": 0.9186548793797014, "nbUnion": 35362, "duplicates": 205, "invented": 88 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 482 }, "info": { "protonNMR": { "present": 596, "notPresent": 46, "valid": 238, "notValid": 358 }, "carbonNMR": { "present": 439, "notPresent": 203, "valid": 174, "notValid": 265 }, "elementalAnalysis": { "present": 259, "notPresent": 383, "valid": 217, "notValid": 42 }, "hrms": { "present": 76, "notPresent": 566, "valid": 12, "notValid": 64 } }, "logs": { "info": 81, "warn": 194, "error": 207, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449445, "totalTokens": 969196, "nbPrompts": 642, "version": "v107", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.418641 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v107" }, { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v108.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v108" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Our baseline prompt. Replicate 8" }, "stats": { "nmr": { "proton": { "nbSpectra": 592, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 22, "500": 5 }, "solvent": { "CDCl3": 95, "not specified": 13, "CD3OD": 32, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 6, "null": 1, "CDC13": 1, "N/A": 1, "literature": 1, "pyridine-d5": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 437, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 29, "CDCl3": 77, "MeOH": 5, "null": 1, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 7, "not specified": 10, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2039 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 0, "counter": 6 }, { "number": 15, "counter": 4 }, { "number": 400, "counter": 4 }, { "number": 29, "counter": 2 }, { "number": 32, "counter": 2 }, { "number": 37, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 8, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 77, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 5.42, "counter": 1 }, { "number": 5.56, "counter": 1 }, { "number": 18.89, "counter": 1 }, { "number": 19.03, "counter": 1 }, { "number": 60.95, "counter": 1 }, { "number": 61.12, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 159 }, { "number": 1, "counter": 94 }, { "number": 14, "counter": 47 }, { "number": 10, "counter": 42 }, { "number": 16, "counter": 41 }, { "number": 8, "counter": 38 }, { "number": 18, "counter": 36 }, { "number": 9, "counter": 35 }, { "number": 4, "counter": 33 }, { "number": 12, "counter": 33 }, { "number": 15, "counter": 32 }, { "number": 6, "counter": 31 }, { "number": 5, "counter": 29 }, { "number": 17, "counter": 29 }, { "number": 3, "counter": 27 }, { "number": 2, "counter": 25 }, { "number": 25, "counter": 21 }, { "number": 20, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 11, "counter": 19 }, { "number": 24, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 23, "counter": 18 }, { "number": 21, "counter": 17 }, { "number": 28, "counter": 16 }, { "number": 100, "counter": 13 }, { "number": 22, "counter": 10 }, { "number": 27, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": -10, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": 282, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": 196, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 31, "counter": 5 }, { "number": 32, "counter": 5 }, { "number": 43, "counter": 5 }, { "number": 48, "counter": 5 } ], "valid": 526, "invalid": 116, "ratioPresent": 0.9204248092301613, "nbUnion": 35375, "duplicates": 228, "invented": 98 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 481 }, "info": { "protonNMR": { "present": 593, "notPresent": 49, "valid": 231, "notValid": 362 }, "carbonNMR": { "present": 437, "notPresent": 205, "valid": 174, "notValid": 263 }, "elementalAnalysis": { "present": 260, "notPresent": 382, "valid": 214, "notValid": 46 }, "hrms": { "present": 85, "notPresent": 557, "valid": 12, "notValid": 73 } }, "logs": { "info": 76, "warn": 207, "error": 198, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447198, "totalTokens": 966949, "nbPrompts": 642, "version": "v108", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.4141469999999998 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.2_0", "version": "v108" } ], "versions": [ "v1", "v101", "v102", "v103", "v104", "v105", "v106", "v107", "v108" ], "stats": { "yaml": { "parsable": [ 640, 640, 642, 641, 639, 640, 641, 639, 641 ], "notParsable": [ 2, 2, 0, 1, 3, 2, 1, 3, 1 ], "empty": [ 1986, 1826, 1982, 2416, 2208, 1874, 2133, 2221, 2039 ] }, "numbers": { "valid": [ 529, 527, 542, 539, 527, 530, 533, 530, 526 ], "invalid": [ 113, 115, 100, 103, 115, 112, 109, 112, 116 ], "ratioPresent": [ 0.9157177263296138, 0.9194799843169938, 0.9245445611070909, 0.9219121270865928, 0.9160045890062518, 0.9178019253763291, 0.9176873790725858, 0.9186548793797014, 0.9204248092301613 ], "nbUnion": [ 35384, 35378, 35390, 35365, 35383, 35398, 35393, 35362, 35375 ], "duplicates": [ 199, 234, 253, 194, 215, 224, 206, 205, 228 ], "invented": [ 103, 103, 109, 88, 107, 121, 121, 88, 98 ] }, "meta": { "withName": [ 637, 638, 640, 639, 636, 636, 639, 638, 639 ], "withoutName": [ 5, 4, 2, 3, 6, 6, 3, 4, 3 ], "withMolfile": [ 481, 484, 483, 481, 481, 480, 483, 482, 481 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.2_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 591, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 67, "500": 6, "N/A": 1 }, "solvent": { "N/A": 25, "CDCl3": 130, "not specified": 37, "unknown": 31, "uncertain": 2, "CD3OD": 32, "Not specified": 19, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "MeOH": 11, "CDC13": 1, "literature": 3, "pyridine-d5": 7, "MeOD": 3, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "CHCl3-EtOH (1:1)": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 498, "frequency": { "50": 2, "75": 5, "80": 2, "100": 62, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 56, "unknown": 27, "CDCl3": 117, "uncertain": 2, "not specified": 27, "CD3OD": 30, "N/A": 14, "Not specified": 14, "CD3COCD3": 2, "null": 1, "MeOH": 5, "literature": 3, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 13644 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 131 }, { "number": 400, "counter": 49 }, { "number": 100, "counter": 23 }, { "number": -18, "counter": 2 }, { "number": 12, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 148 }, { "number": 1, "counter": 56 }, { "number": 100, "counter": 41 }, { "number": 14, "counter": 36 }, { "number": 15, "counter": 33 }, { "number": 4, "counter": 31 }, { "number": 18, "counter": 29 }, { "number": 8, "counter": 27 }, { "number": 10, "counter": 27 }, { "number": 17, "counter": 26 }, { "number": 9, "counter": 25 }, { "number": 6, "counter": 24 }, { "number": 12, "counter": 24 }, { "number": 16, "counter": 24 }, { "number": 5, "counter": 22 }, { "number": 7, "counter": 22 }, { "number": 2, "counter": 21 }, { "number": 24, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 3, "counter": 18 }, { "number": 224, "counter": 18 }, { "number": 211, "counter": 17 }, { "number": 19, "counter": 16 }, { "number": 20, "counter": 16 }, { "number": 239, "counter": 16 }, { "number": 11, "counter": 15 }, { "number": 21, "counter": 15 }, { "number": 196, "counter": 15 }, { "number": 267, "counter": 15 }, { "number": 28, "counter": 13 }, { "number": 32, "counter": 13 }, { "number": 23, "counter": 12 }, { "number": 254, "counter": 12 }, { "number": 282, "counter": 12 }, { "number": 27, "counter": 11 }, { "number": 70, "counter": 11 }, { "number": -8, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": 43, "counter": 10 }, { "number": 135, "counter": 10 }, { "number": 30, "counter": 9 }, { "number": 225, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": -13, "counter": 8 }, { "number": -12, "counter": 8 }, { "number": -5, "counter": 8 }, { "number": 31, "counter": 7 }, { "number": 33, "counter": 7 } ], "valid": 407, "invalid": 235, "ratioPresent": 0.9113576290371678, "nbUnion": 35516, "duplicates": 148, "invented": 500 }, "meta": { "withName": 634, "withoutName": 8, "withMolfile": 481 }, "info": { "protonNMR": { "present": 607, "notPresent": 35, "valid": 168, "notValid": 439 }, "carbonNMR": { "present": 509, "notPresent": 133, "valid": 144, "notValid": 365 }, "elementalAnalysis": { "present": 271, "notPresent": 371, "valid": 222, "notValid": 49 }, "hrms": { "present": 63, "notPresent": 579, "valid": 11, "notValid": 52 } }, "logs": { "info": 40, "warn": 179, "error": 262, "fatal": 161 }, "chatgpt": { "promptTokens": 517825, "completionTokens": 508628, "totalTokens": 1026453, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.535081 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.2_0", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 639 ], "notParsable": [ 3 ], "empty": [ 13644 ] }, "numbers": { "valid": [ 407 ], "invalid": [ 235 ], "ratioPresent": [ 0.9113576290371678 ], "nbUnion": [ 35516 ], "duplicates": [ 148 ], "invented": [ 500 ] }, "meta": { "withName": [ 634 ], "withoutName": [ 8 ], "withMolfile": [ 481 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.2_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 600, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 63, "500": 4, "7.39": 1, "Not specified": 1 }, "solvent": { "CDCl3": 124, "not specified": 27, "DMSO-d6": 73, "CD3OD": 33, "δ/ppm": 1, "unknown": 5, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "MeOH": 7, "dioxane": 1, "CDC13": 1, "N/A": 9, "literature": 3, "pyridine-d5": 7, "Not specified": 11, "MeOD": 3, "dimethylformamide": 1, "D2O": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 6, "acetone-d6": 4, "unspecified": 2 } }, "carbon": { "nbSpectra": 449, "frequency": { "50": 2, "75": 5, "80": 2, "100": 60, "125": 3, "200": 2, "100.6": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 106, "not specified": 13, "CD3OD": 32, "unknown": 6, "null": 1, "CD3COCD3": 1, "N/A": 9, "literature": 3, "pyridine-d5": 7, "MeOH": 6, "Not specified": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "unspecified": 3, "acetone-d6": 5 } } }, "yaml": { "parsable": 637, "notParsable": 5, "empty": 4471 }, "numbers": { "inventedNumbers": [ { "number": 400, "counter": 45 }, { "number": 100, "counter": 34 }, { "number": 0, "counter": 16 }, { "number": 15, "counter": 3 }, { "number": 50, "counter": 3 }, { "number": 70, "counter": 3 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 11, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 47, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.28, "counter": 1 }, { "number": 3.34, "counter": 1 }, { "number": 7.59, "counter": 1 }, { "number": 7.63, "counter": 1 }, { "number": 16.62, "counter": 1 }, { "number": 16.78, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 431 }, { "number": -1, "counter": 139 }, { "number": 1, "counter": 60 }, { "number": 14, "counter": 56 }, { "number": 16, "counter": 41 }, { "number": 10, "counter": 36 }, { "number": 15, "counter": 36 }, { "number": 6, "counter": 34 }, { "number": 5, "counter": 31 }, { "number": 9, "counter": 31 }, { "number": 17, "counter": 31 }, { "number": 18, "counter": 31 }, { "number": 3, "counter": 30 }, { "number": 12, "counter": 30 }, { "number": 4, "counter": 29 }, { "number": 8, "counter": 25 }, { "number": 24, "counter": 23 }, { "number": 23, "counter": 22 }, { "number": 25, "counter": 21 }, { "number": 2, "counter": 20 }, { "number": 19, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 20, "counter": 17 }, { "number": 7, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 22, "counter": 13 }, { "number": 28, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": 100, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 36, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -7, "counter": 7 }, { "number": 31, "counter": 6 }, { "number": 224, "counter": 6 }, { "number": 253, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -13, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 197, "counter": 5 }, { "number": 211, "counter": 5 } ], "valid": 458, "invalid": 184, "ratioPresent": 0.9266622911696794, "nbUnion": 35438, "duplicates": 331, "invented": 185 }, "meta": { "withName": 635, "withoutName": 7, "withMolfile": 478 }, "info": { "protonNMR": { "present": 602, "notPresent": 40, "valid": 220, "notValid": 382 }, "carbonNMR": { "present": 451, "notPresent": 191, "valid": 163, "notValid": 288 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 221, "notValid": 43 }, "hrms": { "present": 88, "notPresent": 554, "valid": 13, "notValid": 75 } }, "logs": { "info": 60, "warn": 202, "error": 216, "fatal": 164 }, "chatgpt": { "promptTokens": 507553, "completionTokens": 478348, "totalTokens": 985901, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.4642490000000001 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.2_0", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 637 ], "notParsable": [ 5 ], "empty": [ 4471 ] }, "numbers": { "valid": [ 458 ], "invalid": [ 184 ], "ratioPresent": [ 0.9266622911696794 ], "nbUnion": [ 35438 ], "duplicates": [ 331 ], "invented": [ 185 ] }, "meta": { "withName": [ 635 ], "withoutName": [ 7 ], "withMolfile": [ 478 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_0.2_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 542, "frequency": { "50": 4, "80": 2, "200": 2, "300": 13, "400": 23, "500": 5, "not specified": 4, "Not specified": 1 }, "solvent": { "CDCl3": 92, "not specified": 34, "DMSO-d6": 70, "uncertain": 5, "CD3OD": 29, "Not specified": 15, "unspecified": 2, "unknown": 8, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 5, "null": 2, "CDC13": 1, "pyridine-d5": 6, "MeOD": 3, "D2O": 2, "d4-MeOH": 30, "CDCl3 + d4-MeOH": 1, "DMSO": 1, "see Table 2": 1, "acetone-d6": 3, "not provided": 2, "δ": 2, "CH3OH": 1 } }, "carbon": { "nbSpectra": 413, "frequency": { "50": 2, "75": 7, "80": 2, "100": 23, "125": 3, "50.3": 1, "75.4": 1, "not specified": 4 }, "solvent": { "DMSO-d6": 48, "not specified": 30, "uncertain": 2, "CDCl3": 77, "CD3OD": 29, "Not specified": 11, "unspecified": 2, "unknown": 7, "CD3COCD3": 1, "pyridine-d5": 5, "MeOH": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 28, "CDCl3 + d4-MeOH": 1, "acetone-d6": 5, "null": 1, "not provided": 2, "δ": 2, "CH3OH": 1 } } }, "yaml": { "parsable": 581, "notParsable": 61, "empty": 4185 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 11 }, { "number": 0, "counter": 9 }, { "number": 400, "counter": 7 }, { "number": 50, "counter": 3 }, { "number": 75, "counter": 2 }, { "number": 300, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": -3, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 424 }, { "number": -1, "counter": 200 }, { "number": 1, "counter": 141 }, { "number": 3, "counter": 88 }, { "number": 2, "counter": 82 }, { "number": 4, "counter": 75 }, { "number": 6, "counter": 72 }, { "number": 5, "counter": 60 }, { "number": 14, "counter": 54 }, { "number": 8, "counter": 47 }, { "number": 15, "counter": 45 }, { "number": 10, "counter": 42 }, { "number": 7, "counter": 38 }, { "number": 12, "counter": 37 }, { "number": 18, "counter": 37 }, { "number": 16, "counter": 36 }, { "number": 9, "counter": 35 }, { "number": 100, "counter": 35 }, { "number": 11, "counter": 29 }, { "number": -4, "counter": 29 }, { "number": 17, "counter": 27 }, { "number": -3, "counter": 26 }, { "number": 24, "counter": 25 }, { "number": 20, "counter": 23 }, { "number": -2, "counter": 23 }, { "number": 25, "counter": 21 }, { "number": 23, "counter": 19 }, { "number": 21, "counter": 18 }, { "number": -5, "counter": 18 }, { "number": 22, "counter": 17 }, { "number": 19, "counter": 16 }, { "number": -7, "counter": 15 }, { "number": 28, "counter": 14 }, { "number": -6, "counter": 14 }, { "number": -8, "counter": 14 }, { "number": 27, "counter": 11 }, { "number": 70, "counter": 11 }, { "number": 7.28, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": -11, "counter": 10 }, { "number": 7.3, "counter": 10 }, { "number": 32, "counter": 9 }, { "number": 63, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 7.5, "counter": 9 }, { "number": 7.9, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 43, "counter": 8 } ], "valid": 509, "invalid": 133, "ratioPresent": 0.8324272045399188, "nbUnion": 35349, "duplicates": 460, "invented": 85 }, "meta": { "withName": 571, "withoutName": 71, "withMolfile": 435 }, "info": { "protonNMR": { "present": 542, "notPresent": 100, "valid": 178, "notValid": 364 }, "carbonNMR": { "present": 413, "notPresent": 229, "valid": 146, "notValid": 267 }, "elementalAnalysis": { "present": 231, "notPresent": 411, "valid": 188, "notValid": 43 }, "hrms": { "present": 85, "notPresent": 557, "valid": 14, "notValid": 71 } }, "logs": { "info": 59, "warn": 167, "error": 209, "fatal": 207 }, "chatgpt": { "promptTokens": 504985, "completionTokens": 484910, "totalTokens": 989895, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.474805 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_0.2_0", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 581 ], "notParsable": [ 61 ], "empty": [ 4185 ] }, "numbers": { "valid": [ 509 ], "invalid": [ 133 ], "ratioPresent": [ 0.8324272045399188 ], "nbUnion": [ 35349 ], "duplicates": [ 460 ], "invented": [ 85 ] }, "meta": { "withName": [ 571 ], "withoutName": [ 71 ], "withMolfile": [ 435 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_0.2_0", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.2-0_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.2-0_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 590, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 5 }, "solvent": { "CDCl3": 97, "not specified": 7, "ethanol-water": 1, "CD3OD": 32, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "unknown": 1, "N/A": 3, "literature": 1, "pyridine-d5": 7, "MeOH": 9, "Not specified": 3, "MeOD": 3, "unspecified": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "water/methanol": 1, "d": 3, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 440, "frequency": { "50": 2, "75": 5, "80": 2, "100": 23, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 79, "CD3OD": 29, "null": 1, "unknown": 1, "CD3COCD3": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 9, "not specified": 4, "Not specified": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2307 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 400, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 50, "counter": 4 }, { "number": 15, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 8.5, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 5, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.88, "counter": 1 }, { "number": 5.32, "counter": 1 }, { "number": 11.17, "counter": 1 }, { "number": 64.63, "counter": 1 }, { "number": 5.05, "counter": 1 }, { "number": 5.12, "counter": 1 }, { "number": 16.4, "counter": 1 }, { "number": 16.42, "counter": 1 }, { "number": 17.68, "counter": 1 }, { "number": 17.81, "counter": 1 }, { "number": 60.65, "counter": 1 }, { "number": 60.87, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 430 }, { "number": -1, "counter": 166 }, { "number": 1, "counter": 75 }, { "number": 14, "counter": 53 }, { "number": 10, "counter": 44 }, { "number": 15, "counter": 43 }, { "number": 8, "counter": 40 }, { "number": 12, "counter": 38 }, { "number": 18, "counter": 36 }, { "number": 9, "counter": 32 }, { "number": 16, "counter": 32 }, { "number": 5, "counter": 30 }, { "number": 4, "counter": 28 }, { "number": 3, "counter": 27 }, { "number": 6, "counter": 27 }, { "number": 17, "counter": 25 }, { "number": 2, "counter": 24 }, { "number": 7, "counter": 24 }, { "number": 24, "counter": 22 }, { "number": 11, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 20, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 19, "counter": 17 }, { "number": 21, "counter": 16 }, { "number": 28, "counter": 16 }, { "number": 100, "counter": 15 }, { "number": 224, "counter": 12 }, { "number": -10, "counter": 12 }, { "number": -9, "counter": 12 }, { "number": -4, "counter": 12 }, { "number": 22, "counter": 11 }, { "number": 27, "counter": 11 }, { "number": 211, "counter": 11 }, { "number": 239, "counter": 11 }, { "number": 196, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 267, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": -2, "counter": 7 }, { "number": -3, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 34, "counter": 6 }, { "number": 282, "counter": 6 } ], "valid": 514, "invalid": 128, "ratioPresent": 0.9128887794916273, "nbUnion": 35378, "duplicates": 229, "invented": 109 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 479 }, "info": { "protonNMR": { "present": 590, "notPresent": 52, "valid": 236, "notValid": 354 }, "carbonNMR": { "present": 440, "notPresent": 202, "valid": 167, "notValid": 273 }, "elementalAnalysis": { "present": 260, "notPresent": 382, "valid": 214, "notValid": 46 }, "hrms": { "present": 81, "notPresent": 561, "valid": 14, "notValid": 67 } }, "logs": { "info": 73, "warn": 225, "error": 181, "fatal": 163 }, "chatgpt": { "promptTokens": 513331, "completionTokens": 447728, "totalTokens": 961059, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 0.2, "top_p": 0, "cost": 1.408787 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_0.2_0", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 639 ], "notParsable": [ 3 ], "empty": [ 2307 ] }, "numbers": { "valid": [ 514 ], "invalid": [ 128 ], "ratioPresent": [ 0.9128887794916273 ], "nbUnion": [ 35378 ], "duplicates": [ 229 ], "invented": [ 109 ] }, "meta": { "withName": [ 639 ], "withoutName": [ 3 ], "withMolfile": [ 479 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 599, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 20, "500": 5 }, "solvent": { "CDCl3": 94, "unknown": 3, "not provided": 1, "not specified": 5, "CD3OD": 34, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "CDC13": 1, "N/A": 1, "pyridine-d5": 7, "Not specified": 5, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 433, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 49, "CD3OD": 31, "CDCl3": 77, "null": 1, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 6, "Not specified": 4, "MeOH": 7, "not specified": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 1741 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 15, "counter": 3 }, { "number": 22, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 32, "counter": 2 }, { "number": 41, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": 400, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 77.85, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": -23.86, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 2.89, "counter": 1 }, { "number": 8.37, "counter": 1 }, { "number": 10.89, "counter": 1 }, { "number": 2.99, "counter": 1 }, { "number": 9.01, "counter": 1 }, { "number": 10.05, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 164 }, { "number": 1, "counter": 86 }, { "number": 14, "counter": 46 }, { "number": 10, "counter": 42 }, { "number": 15, "counter": 35 }, { "number": 16, "counter": 35 }, { "number": 6, "counter": 34 }, { "number": 18, "counter": 34 }, { "number": 8, "counter": 31 }, { "number": 9, "counter": 31 }, { "number": 12, "counter": 31 }, { "number": 3, "counter": 30 }, { "number": 4, "counter": 30 }, { "number": 5, "counter": 30 }, { "number": 17, "counter": 27 }, { "number": 2, "counter": 23 }, { "number": 24, "counter": 23 }, { "number": 7, "counter": 22 }, { "number": 19, "counter": 22 }, { "number": 23, "counter": 21 }, { "number": 25, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 20, "counter": 17 }, { "number": 28, "counter": 15 }, { "number": 100, "counter": 15 }, { "number": 21, "counter": 14 }, { "number": 22, "counter": 12 }, { "number": 211, "counter": 12 }, { "number": 239, "counter": 12 }, { "number": 224, "counter": 11 }, { "number": 267, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 196, "counter": 10 }, { "number": 282, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": 254, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": -9, "counter": 8 }, { "number": 27, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 237, "counter": 6 }, { "number": 251, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": 32, "counter": 5 }, { "number": 33, "counter": 5 } ], "valid": 531, "invalid": 111, "ratioPresent": 0.9201566316862111, "nbUnion": 35366, "duplicates": 187, "invented": 82 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 484 }, "info": { "protonNMR": { "present": 600, "notPresent": 42, "valid": 236, "notValid": 364 }, "carbonNMR": { "present": 434, "notPresent": 208, "valid": 171, "notValid": 263 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 214, "notValid": 47 }, "hrms": { "present": 87, "notPresent": 555, "valid": 12, "notValid": 75 } }, "logs": { "info": 73, "warn": 203, "error": 208, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 444990, "totalTokens": 964741, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.4097309999999998 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v1" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v101.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v101" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 1" }, "stats": { "nmr": { "proton": { "nbSpectra": 599, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 5 }, "solvent": { "CDCl3": 95, "not specified": 9, "DMSO-d6": 74, "CD3OD": 34, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "unknown": 2, "pyridine-d5": 7, "MeOH": 8, "Not specified": 3, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 442, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CD3OD": 31, "CDCl3": 76, "null": 2, "CD3COCD3": 1, "unknown": 1, "pyridine-d5": 7, "not specified": 6, "MeOH": 6, "Not specified": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2316 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 19 }, { "number": 31, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 400, "counter": 3 }, { "number": 7, "counter": 2 }, { "number": 14, "counter": 2 }, { "number": 15, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 36, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 11, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 58, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 4.38, "counter": 1 }, { "number": 4.4, "counter": 1 }, { "number": 6.83, "counter": 1 }, { "number": 6.95, "counter": 1 }, { "number": 24.12, "counter": 1 }, { "number": 24.35, "counter": 1 }, { "number": 64.42, "counter": 1 }, { "number": 64.55, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 166 }, { "number": 1, "counter": 93 }, { "number": 14, "counter": 46 }, { "number": 10, "counter": 43 }, { "number": 16, "counter": 37 }, { "number": 18, "counter": 36 }, { "number": 8, "counter": 35 }, { "number": 15, "counter": 35 }, { "number": 9, "counter": 34 }, { "number": 12, "counter": 34 }, { "number": 2, "counter": 31 }, { "number": 4, "counter": 31 }, { "number": 5, "counter": 30 }, { "number": 3, "counter": 28 }, { "number": 6, "counter": 28 }, { "number": 17, "counter": 24 }, { "number": 11, "counter": 22 }, { "number": 24, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 7, "counter": 17 }, { "number": 25, "counter": 17 }, { "number": 21, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 20, "counter": 12 }, { "number": 100, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": 239, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": 26, "counter": 10 }, { "number": 27, "counter": 10 }, { "number": 211, "counter": 10 }, { "number": 196, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 267, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -10, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 31, "counter": 5 } ], "valid": 516, "invalid": 126, "ratioPresent": 0.9242600850401974, "nbUnion": 35385, "duplicates": 284, "invented": 113 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 484 }, "info": { "protonNMR": { "present": 601, "notPresent": 41, "valid": 234, "notValid": 367 }, "carbonNMR": { "present": 442, "notPresent": 200, "valid": 174, "notValid": 268 }, "elementalAnalysis": { "present": 266, "notPresent": 376, "valid": 221, "notValid": 45 }, "hrms": { "present": 88, "notPresent": 554, "valid": 14, "notValid": 74 } }, "logs": { "info": 79, "warn": 197, "error": 208, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 451100, "totalTokens": 970851, "nbPrompts": 642, "version": "v101", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.421951 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v101" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v102.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v102" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 2." }, "stats": { "nmr": { "proton": { "nbSpectra": 597, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 5 }, "solvent": { "CDCl3": 94, "not specified": 9, "DMSO-d6": 74, "CD3OD": 33, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "N/A": 2, "unknown": 1, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "Not specified": 1 } }, "carbon": { "nbSpectra": 441, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 30, "null": 2, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 6, "not specified": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2144 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 20 }, { "number": 0, "counter": 6 }, { "number": 50, "counter": 3 }, { "number": 400, "counter": 3 }, { "number": 15, "counter": 2 }, { "number": 28, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 19, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.31, "counter": 1 }, { "number": 7.53, "counter": 1 }, { "number": 7.68, "counter": 1 }, { "number": 28.29, "counter": 1 }, { "number": 28.67, "counter": 1 }, { "number": 60.65, "counter": 1 }, { "number": 60.87, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.21, "counter": 1 }, { "number": 3.25, "counter": 1 }, { "number": 6.79, "counter": 1 }, { "number": 6.81, "counter": 1 }, { "number": 12.18, "counter": 1 }, { "number": 12.33, "counter": 1 }, { "number": 77.65, "counter": 1 }, { "number": 77.78, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 429 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 90 }, { "number": 14, "counter": 50 }, { "number": 10, "counter": 42 }, { "number": 6, "counter": 38 }, { "number": 8, "counter": 37 }, { "number": 9, "counter": 36 }, { "number": 5, "counter": 34 }, { "number": 12, "counter": 34 }, { "number": 15, "counter": 34 }, { "number": 3, "counter": 32 }, { "number": 18, "counter": 32 }, { "number": 4, "counter": 31 }, { "number": 16, "counter": 31 }, { "number": 2, "counter": 29 }, { "number": 17, "counter": 26 }, { "number": 23, "counter": 22 }, { "number": 25, "counter": 22 }, { "number": 24, "counter": 21 }, { "number": 11, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 19, "counter": 17 }, { "number": 20, "counter": 16 }, { "number": 21, "counter": 16 }, { "number": 28, "counter": 15 }, { "number": 22, "counter": 14 }, { "number": 100, "counter": 14 }, { "number": 26, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": 211, "counter": 10 }, { "number": 267, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": 196, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": -9, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": 43, "counter": 5 } ], "valid": 528, "invalid": 114, "ratioPresent": 0.921435546700276, "nbUnion": 35362, "duplicates": 236, "invented": 91 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 484 }, "info": { "protonNMR": { "present": 598, "notPresent": 44, "valid": 236, "notValid": 362 }, "carbonNMR": { "present": 441, "notPresent": 201, "valid": 173, "notValid": 268 }, "elementalAnalysis": { "present": 265, "notPresent": 377, "valid": 219, "notValid": 46 }, "hrms": { "present": 86, "notPresent": 556, "valid": 14, "notValid": 72 } }, "logs": { "info": 77, "warn": 199, "error": 208, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447730, "totalTokens": 967481, "nbPrompts": 642, "version": "v102", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.415211 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v102" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v103.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v103" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 3" }, "stats": { "nmr": { "proton": { "nbSpectra": 592, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 5, "not specified": 1 }, "solvent": { "CDCl3": 96, "not specified": 8, "CD3OD": 32, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 6, "null": 1, "CDC13": 1, "unknown": 2, "pyridine-d5": 7, "Not specified": 3, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 440, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "not specified": 4, "CD3OD": 29, "CDCl3": 78, "null": 1, "CD3COCD3": 1, "pyridine-d5": 7, "MeOH": 4, "Not specified": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "unknown": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2180 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 15 }, { "number": 400, "counter": 6 }, { "number": 33, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 28, "counter": 3 }, { "number": 15, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 11, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1600, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3400, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 3.49, "counter": 1 }, { "number": 7.81, "counter": 1 }, { "number": 10.83, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 425 }, { "number": -1, "counter": 154 }, { "number": 1, "counter": 87 }, { "number": 14, "counter": 50 }, { "number": 10, "counter": 42 }, { "number": 16, "counter": 39 }, { "number": 8, "counter": 35 }, { "number": 9, "counter": 34 }, { "number": 15, "counter": 34 }, { "number": 12, "counter": 33 }, { "number": 6, "counter": 32 }, { "number": 5, "counter": 30 }, { "number": 2, "counter": 29 }, { "number": 18, "counter": 28 }, { "number": 4, "counter": 27 }, { "number": 17, "counter": 27 }, { "number": 3, "counter": 24 }, { "number": 11, "counter": 21 }, { "number": 20, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 24, "counter": 19 }, { "number": 21, "counter": 17 }, { "number": 7, "counter": 16 }, { "number": 19, "counter": 16 }, { "number": 25, "counter": 16 }, { "number": 22, "counter": 13 }, { "number": 224, "counter": 13 }, { "number": 26, "counter": 11 }, { "number": 100, "counter": 11 }, { "number": 28, "counter": 10 }, { "number": 211, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 196, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -9, "counter": 8 }, { "number": 225, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 31, "counter": 5 }, { "number": 32, "counter": 5 }, { "number": 48, "counter": 5 }, { "number": 0.1, "counter": 5 } ], "valid": 526, "invalid": 116, "ratioPresent": 0.9246131459935621, "nbUnion": 35367, "duplicates": 230, "invented": 90 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 483 }, "info": { "protonNMR": { "present": 593, "notPresent": 49, "valid": 231, "notValid": 362 }, "carbonNMR": { "present": 441, "notPresent": 201, "valid": 173, "notValid": 268 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 220, "notValid": 44 }, "hrms": { "present": 81, "notPresent": 561, "valid": 14, "notValid": 67 } }, "logs": { "info": 77, "warn": 210, "error": 196, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 451580, "totalTokens": 971331, "nbPrompts": 642, "version": "v103", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.422911 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v103" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v104.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v104" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 4" }, "stats": { "nmr": { "proton": { "nbSpectra": 595, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 5, "Not specified": 1 }, "solvent": { "CDCl3": 96, "not specified": 10, "CD3OD": 32, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 6, "null": 2, "unknown": 2, "CDC13": 1, "Not specified": 4, "pyridine-d5": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "N/A": 1, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 438, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "Not specified": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 77, "CD3OD": 30, "null": 2, "unknown": 1, "CD3COCD3": 1, "Not specified": 4, "not specified": 9, "pyridine-d5": 7, "MeOH": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "N/A": 1, "acetone-d6": 4 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2184 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 15 }, { "number": 400, "counter": 5 }, { "number": 0, "counter": 4 }, { "number": 15, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 39, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 1.68, "counter": 1 }, { "number": 1.86, "counter": 1 }, { "number": 8.63, "counter": 1 }, { "number": 8.72, "counter": 1 }, { "number": 28.53, "counter": 1 }, { "number": 28.78, "counter": 1 }, { "number": 60.82, "counter": 1 }, { "number": 60.98, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 164 }, { "number": 1, "counter": 95 }, { "number": 14, "counter": 49 }, { "number": 10, "counter": 46 }, { "number": 16, "counter": 41 }, { "number": 15, "counter": 39 }, { "number": 12, "counter": 37 }, { "number": 4, "counter": 35 }, { "number": 8, "counter": 35 }, { "number": 18, "counter": 35 }, { "number": 6, "counter": 33 }, { "number": 9, "counter": 32 }, { "number": 5, "counter": 30 }, { "number": 2, "counter": 29 }, { "number": 17, "counter": 27 }, { "number": 3, "counter": 25 }, { "number": 7, "counter": 22 }, { "number": 11, "counter": 22 }, { "number": 23, "counter": 21 }, { "number": 24, "counter": 21 }, { "number": 20, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 21, "counter": 18 }, { "number": 25, "counter": 17 }, { "number": 28, "counter": 16 }, { "number": 22, "counter": 12 }, { "number": 100, "counter": 12 }, { "number": 224, "counter": 11 }, { "number": 211, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": 267, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 27, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 196, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -9, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -13, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -8, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": -12, "counter": 6 } ], "valid": 535, "invalid": 107, "ratioPresent": 0.9212136864348677, "nbUnion": 35380, "duplicates": 195, "invented": 107 }, "meta": { "withName": 640, "withoutName": 2, "withMolfile": 485 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 236, "notValid": 361 }, "carbonNMR": { "present": 439, "notPresent": 203, "valid": 166, "notValid": 273 }, "elementalAnalysis": { "present": 267, "notPresent": 375, "valid": 216, "notValid": 51 }, "hrms": { "present": 85, "notPresent": 557, "valid": 13, "notValid": 72 } }, "logs": { "info": 70, "warn": 213, "error": 202, "fatal": 157 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448575, "totalTokens": 968326, "nbPrompts": 642, "version": "v104", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.416901 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v104" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v105.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v105" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 5 }, "solvent": { "CDCl3": 97, "not specified": 9, "CD3OD": 35, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unspecified": 1, "null": 1, "CDC13": 1, "N/A": 1, "Not specified": 3, "unknown": 2, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 445, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CD3OD": 32, "CDCl3": 76, "unspecified": 1, "null": 1, "CD3COCD3": 1, "N/A": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 4, "not specified": 7, "TBD": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2396 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 12 }, { "number": 400, "counter": 6 }, { "number": 0, "counter": 4 }, { "number": 15, "counter": 4 }, { "number": 28, "counter": 4 }, { "number": 26, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 33, "counter": 2 }, { "number": 37, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 5, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.7, "counter": 1 }, { "number": 3.86, "counter": 1 }, { "number": 10.82, "counter": 1 }, { "number": 11.48, "counter": 1 }, { "number": 19.65, "counter": 1 }, { "number": 20.34, "counter": 1 }, { "number": 64.98, "counter": 1 }, { "number": 65.17, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 430 }, { "number": -1, "counter": 166 }, { "number": 1, "counter": 88 }, { "number": 14, "counter": 45 }, { "number": 10, "counter": 42 }, { "number": 16, "counter": 38 }, { "number": 18, "counter": 37 }, { "number": 15, "counter": 36 }, { "number": 8, "counter": 35 }, { "number": 9, "counter": 35 }, { "number": 12, "counter": 35 }, { "number": 6, "counter": 33 }, { "number": 3, "counter": 31 }, { "number": 5, "counter": 30 }, { "number": 4, "counter": 28 }, { "number": 2, "counter": 27 }, { "number": 17, "counter": 25 }, { "number": 24, "counter": 24 }, { "number": 11, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 19, "counter": 19 }, { "number": 20, "counter": 19 }, { "number": 21, "counter": 18 }, { "number": 23, "counter": 18 }, { "number": 28, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": -7, "counter": 11 }, { "number": 211, "counter": 10 }, { "number": -10, "counter": 10 }, { "number": 224, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 27, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": 30, "counter": 5 }, { "number": 230, "counter": 5 }, { "number": 254, "counter": 5 }, { "number": 277, "counter": 5 }, { "number": 338, "counter": 5 } ], "valid": 532, "invalid": 110, "ratioPresent": 0.9210625173852902, "nbUnion": 35379, "duplicates": 173, "invented": 105 }, "meta": { "withName": 640, "withoutName": 2, "withMolfile": 484 }, "info": { "protonNMR": { "present": 598, "notPresent": 44, "valid": 238, "notValid": 360 }, "carbonNMR": { "present": 447, "notPresent": 195, "valid": 175, "notValid": 272 }, "elementalAnalysis": { "present": 263, "notPresent": 379, "valid": 217, "notValid": 46 }, "hrms": { "present": 80, "notPresent": 562, "valid": 12, "notValid": 68 } }, "logs": { "info": 78, "warn": 209, "error": 197, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450070, "totalTokens": 969821, "nbPrompts": 642, "version": "v105", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.419891 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v105" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v106.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v106" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 6 }, "solvent": { "CDCl3": 96, "not specified": 10, "ethanol-water": 1, "CD3OD": 35, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 3, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 7, "Not specified": 1, "MeOD": 3, "unknown": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 441, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "100.6": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 32, "null": 2, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "not specified": 5, "MeOH": 6, "Not specified": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "unknown": 1 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2056 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 14 }, { "number": 0, "counter": 5 }, { "number": 400, "counter": 5 }, { "number": 15, "counter": 2 }, { "number": 27, "counter": 2 }, { "number": 33, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 20, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 64, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 123, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 498, "counter": 1 }, { "number": 500, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 4.05, "counter": 1 }, { "number": 23.37, "counter": 1 }, { "number": 72.58, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.29, "counter": 1 }, { "number": 8.28, "counter": 1 }, { "number": 8.49, "counter": 1 }, { "number": 79.94, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 3.39, "counter": 1 }, { "number": 7.92, "counter": 1 }, { "number": 11.4, "counter": 1 }, { "number": 77.29, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 422 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 92 }, { "number": 14, "counter": 51 }, { "number": 8, "counter": 40 }, { "number": 10, "counter": 39 }, { "number": 6, "counter": 37 }, { "number": 4, "counter": 36 }, { "number": 15, "counter": 36 }, { "number": 9, "counter": 35 }, { "number": 12, "counter": 35 }, { "number": 16, "counter": 35 }, { "number": 5, "counter": 34 }, { "number": 3, "counter": 33 }, { "number": 2, "counter": 31 }, { "number": 18, "counter": 31 }, { "number": 17, "counter": 28 }, { "number": 24, "counter": 26 }, { "number": 7, "counter": 22 }, { "number": 11, "counter": 22 }, { "number": 19, "counter": 22 }, { "number": 25, "counter": 22 }, { "number": 23, "counter": 19 }, { "number": 20, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 22, "counter": 15 }, { "number": 28, "counter": 15 }, { "number": 100, "counter": 14 }, { "number": 27, "counter": 12 }, { "number": 224, "counter": 12 }, { "number": -7, "counter": 12 }, { "number": -10, "counter": 10 }, { "number": -8, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": 282, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 30, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 33, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 254, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -3, "counter": 6 } ], "valid": 532, "invalid": 110, "ratioPresent": 0.9180902223088676, "nbUnion": 35363, "duplicates": 184, "invented": 94 }, "meta": { "withName": 637, "withoutName": 5, "withMolfile": 481 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 230, "notValid": 367 }, "carbonNMR": { "present": 442, "notPresent": 200, "valid": 172, "notValid": 270 }, "elementalAnalysis": { "present": 257, "notPresent": 385, "valid": 208, "notValid": 49 }, "hrms": { "present": 87, "notPresent": 555, "valid": 12, "notValid": 75 } }, "logs": { "info": 75, "warn": 195, "error": 211, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447741, "totalTokens": 967492, "nbPrompts": 642, "version": "v106", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.415233 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v106" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v107.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v107" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 7" }, "stats": { "nmr": { "proton": { "nbSpectra": 594, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 5 }, "solvent": { "CDCl3": 95, "not specified": 12, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unknown": 3, "null": 2, "CDC13": 1, "N/A": 2, "literature": 1, "pyridine-d5": 7, "MeOH": 4, "Not specified": 1, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 442, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 79, "CD3OD": 30, "unknown": 1, "null": 1, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 3, "not specified": 7, "Not specified": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2065 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 11 }, { "number": 0, "counter": 8 }, { "number": 400, "counter": 5 }, { "number": 15, "counter": 3 }, { "number": 22, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": 3.77, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 45, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.88, "counter": 1 }, { "number": 5.06, "counter": 1 }, { "number": 5.45, "counter": 1 }, { "number": 20.98, "counter": 1 }, { "number": 64.63, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 3.87, "counter": 1 }, { "number": 7.48, "counter": 1 }, { "number": 7.56, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 431 }, { "number": -1, "counter": 161 }, { "number": 1, "counter": 90 }, { "number": 10, "counter": 45 }, { "number": 14, "counter": 45 }, { "number": 16, "counter": 37 }, { "number": 3, "counter": 35 }, { "number": 4, "counter": 35 }, { "number": 15, "counter": 35 }, { "number": 18, "counter": 33 }, { "number": 5, "counter": 32 }, { "number": 8, "counter": 32 }, { "number": 12, "counter": 31 }, { "number": 6, "counter": 30 }, { "number": 9, "counter": 29 }, { "number": 17, "counter": 29 }, { "number": 2, "counter": 28 }, { "number": 7, "counter": 21 }, { "number": 11, "counter": 19 }, { "number": 19, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 24, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 20, "counter": 16 }, { "number": 21, "counter": 13 }, { "number": 28, "counter": 13 }, { "number": 100, "counter": 13 }, { "number": 22, "counter": 10 }, { "number": 27, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 70, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 31, "counter": 7 }, { "number": 43, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 26, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": -14, "counter": 6 }, { "number": -2, "counter": 6 } ], "valid": 528, "invalid": 114, "ratioPresent": 0.9175972983181838, "nbUnion": 35383, "duplicates": 233, "invented": 114 }, "meta": { "withName": 640, "withoutName": 2, "withMolfile": 484 }, "info": { "protonNMR": { "present": 595, "notPresent": 47, "valid": 234, "notValid": 361 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 175, "notValid": 268 }, "elementalAnalysis": { "present": 262, "notPresent": 380, "valid": 216, "notValid": 46 }, "hrms": { "present": 84, "notPresent": 558, "valid": 14, "notValid": 70 } }, "logs": { "info": 81, "warn": 195, "error": 208, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 446454, "totalTokens": 966205, "nbPrompts": 642, "version": "v107", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.412659 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v107" }, { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v108.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v108" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Our baseline prompt. Replicate 8" }, "stats": { "nmr": { "proton": { "nbSpectra": 601, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 5 }, "solvent": { "CDCl3": 95, "not specified": 11, "DMSO-d6": 74, "CD3OD": 35, "CD3COCD3": 2, "CD3CN": 6, "Not specified": 2, "null": 2, "CDC13": 1, "N/A": 1, "unknown": 5, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 442, "frequency": { "50": 2, "75": 5, "80": 2, "100": 23, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 32, "Not specified": 2, "null": 1, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 7, "not specified": 10, "MeOH": 4, "unknown": 2, "TBD": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2521 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 14 }, { "number": 400, "counter": 6 }, { "number": 15, "counter": 3 }, { "number": 30, "counter": 3 }, { "number": 31, "counter": 2 }, { "number": 41, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2.99, "counter": 2 }, { "number": 0, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 77, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 5.07, "counter": 1 }, { "number": 5.23, "counter": 1 }, { "number": 13.47, "counter": 1 }, { "number": 13.67, "counter": 1 }, { "number": 20.18, "counter": 1 }, { "number": 20.31, "counter": 1 }, { "number": 60.95, "counter": 1 }, { "number": 61.12, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 431 }, { "number": -1, "counter": 158 }, { "number": 1, "counter": 90 }, { "number": 14, "counter": 49 }, { "number": 10, "counter": 40 }, { "number": 16, "counter": 37 }, { "number": 8, "counter": 35 }, { "number": 18, "counter": 35 }, { "number": 4, "counter": 34 }, { "number": 9, "counter": 34 }, { "number": 6, "counter": 33 }, { "number": 15, "counter": 33 }, { "number": 3, "counter": 32 }, { "number": 12, "counter": 32 }, { "number": 5, "counter": 30 }, { "number": 17, "counter": 29 }, { "number": 2, "counter": 26 }, { "number": 25, "counter": 22 }, { "number": 24, "counter": 21 }, { "number": 11, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 7, "counter": 17 }, { "number": 20, "counter": 17 }, { "number": 28, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": 21, "counter": 11 }, { "number": 100, "counter": 10 }, { "number": 224, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": 33, "counter": 8 }, { "number": 211, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": -7, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 267, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 36, "counter": 5 }, { "number": 70, "counter": 5 }, { "number": 240, "counter": 5 }, { "number": 254, "counter": 5 } ], "valid": 532, "invalid": 110, "ratioPresent": 0.9230735627713091, "nbUnion": 35377, "duplicates": 272, "invented": 98 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 481 }, "info": { "protonNMR": { "present": 602, "notPresent": 40, "valid": 237, "notValid": 365 }, "carbonNMR": { "present": 444, "notPresent": 198, "valid": 170, "notValid": 274 }, "elementalAnalysis": { "present": 262, "notPresent": 380, "valid": 214, "notValid": 48 }, "hrms": { "present": 78, "notPresent": 564, "valid": 12, "notValid": 66 } }, "logs": { "info": 77, "warn": 206, "error": 198, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 451480, "totalTokens": 971231, "nbPrompts": 642, "version": "v108", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.422711 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.3_0.3", "version": "v108" } ], "versions": [ "v1", "v101", "v102", "v103", "v104", "v105", "v106", "v107", "v108" ], "stats": { "yaml": { "parsable": [ 641, 642, 641, 642, 641, 642, 639, 641, 642 ], "notParsable": [ 1, 0, 1, 0, 1, 0, 3, 1, 0 ], "empty": [ 1741, 2316, 2144, 2180, 2184, 2396, 2056, 2065, 2521 ] }, "numbers": { "valid": [ 531, 516, 528, 526, 535, 532, 532, 528, 532 ], "invalid": [ 111, 126, 114, 116, 107, 110, 110, 114, 110 ], "ratioPresent": [ 0.9201566316862111, 0.9242600850401974, 0.921435546700276, 0.9246131459935621, 0.9212136864348677, 0.9210625173852902, 0.9180902223088676, 0.9175972983181838, 0.9230735627713091 ], "nbUnion": [ 35366, 35385, 35362, 35367, 35380, 35379, 35363, 35383, 35377 ], "duplicates": [ 187, 284, 236, 230, 195, 173, 184, 233, 272 ], "invented": [ 82, 113, 91, 90, 107, 105, 94, 114, 98 ] }, "meta": { "withName": [ 639, 639, 639, 639, 640, 640, 637, 640, 638 ], "withoutName": [ 3, 3, 3, 3, 2, 2, 5, 2, 4 ], "withMolfile": [ 484, 484, 484, 483, 485, 484, 481, 484, 481 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.3_0.3", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 564, "frequency": { "50": 4, "80": 2, "200": 3, "300": 12, "400": 44, "500": 6, "N/A": 1 }, "solvent": { "unknown": 93, "CDCl3": 88, "not specified": 24, "DMSO-d6": 73, "CD3OD": 28, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 7, "CDC13": 1, "N/A": 4, "literature": 1, "pyridine-d5": 7, "Not specified": 2, "ppm": 2, "MeOD": 3, "δ/ppm": 1, "D2O": 4, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "acetone-d6": 2, "δ (ppm)": 1 } }, "carbon": { "nbSpectra": 472, "frequency": { "50": 2, "75": 5, "80": 2, "100": 41, "125": 3, "200": 1, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "unknown": 80, "not specified": 19, "CDCl3": 82, "CD3OD": 26, "CHCl3": 2, "CD3COCD3": 1, "N/A": 4, "literature": 1, "pyridine-d5": 5, "Not specified": 2, "MeOH": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "δ (ppm)": 1, "ppm": 1 } } }, "yaml": { "parsable": 623, "notParsable": 19, "empty": 15204 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 156 }, { "number": 400, "counter": 28 }, { "number": 100, "counter": 11 }, { "number": 30, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 17, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 439 }, { "number": -1, "counter": 147 }, { "number": 1, "counter": 76 }, { "number": 14, "counter": 50 }, { "number": 100, "counter": 44 }, { "number": 4, "counter": 43 }, { "number": 15, "counter": 43 }, { "number": 5, "counter": 42 }, { "number": 18, "counter": 41 }, { "number": 3, "counter": 40 }, { "number": 8, "counter": 39 }, { "number": 12, "counter": 39 }, { "number": 2, "counter": 36 }, { "number": 10, "counter": 36 }, { "number": 16, "counter": 35 }, { "number": 17, "counter": 32 }, { "number": 6, "counter": 29 }, { "number": 9, "counter": 28 }, { "number": 11, "counter": 28 }, { "number": 24, "counter": 24 }, { "number": 25, "counter": 24 }, { "number": 7, "counter": 22 }, { "number": 21, "counter": 20 }, { "number": 19, "counter": 18 }, { "number": 224, "counter": 18 }, { "number": 211, "counter": 16 }, { "number": 239, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": -4, "counter": 16 }, { "number": 20, "counter": 15 }, { "number": 23, "counter": 14 }, { "number": 28, "counter": 14 }, { "number": 30, "counter": 14 }, { "number": 196, "counter": 14 }, { "number": -3, "counter": 14 }, { "number": -10, "counter": 14 }, { "number": -2, "counter": 14 }, { "number": 43, "counter": 13 }, { "number": -8, "counter": 13 }, { "number": 22, "counter": 12 }, { "number": 27, "counter": 12 }, { "number": 135, "counter": 12 }, { "number": -13, "counter": 12 }, { "number": 282, "counter": 11 }, { "number": -7, "counter": 11 }, { "number": -6, "counter": 11 }, { "number": 31, "counter": 10 }, { "number": 32, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": -5, "counter": 10 } ], "valid": 389, "invalid": 253, "ratioPresent": 0.8893666344489312, "nbUnion": 35507, "duplicates": 237, "invented": 596 }, "meta": { "withName": 616, "withoutName": 26, "withMolfile": 468 }, "info": { "protonNMR": { "present": 582, "notPresent": 60, "valid": 160, "notValid": 422 }, "carbonNMR": { "present": 506, "notPresent": 136, "valid": 149, "notValid": 357 }, "elementalAnalysis": { "present": 267, "notPresent": 375, "valid": 221, "notValid": 46 }, "hrms": { "present": 54, "notPresent": 588, "valid": 8, "notValid": 46 } }, "logs": { "info": 36, "warn": 189, "error": 243, "fatal": 174 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 530594, "totalTokens": 1017603, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.548197 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.3_0.3", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 623 ], "notParsable": [ 19 ], "empty": [ 15204 ] }, "numbers": { "valid": [ 389 ], "invalid": [ 253 ], "ratioPresent": [ 0.8893666344489312 ], "nbUnion": [ 35507 ], "duplicates": [ 237 ], "invented": [ 596 ] }, "meta": { "withName": [ 616 ], "withoutName": [ 26 ], "withMolfile": [ 468 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.3_0.3", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 584, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 31, "500": 5 }, "solvent": { "unknown": 9, "not specified": 8, "N/A": 10, "DMSO-d6": 73, "CDCl3": 85, "CD3OD": 32, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 6, "MeOH": 11, "MeOD": 2, "ppm": 1, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 434, "frequency": { "50": 2, "75": 5, "80": 2, "100": 28, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 70, "unknown": 7, "CD3OD": 30, "CHCl3": 1, "CD3COCD3": 1, "N/A": 6, "pyridine-d5": 6, "MeOH": 9, "not specified": 5, "CD3CN": 4, "MeOD": 2, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "CH3OH": 1 } } }, "yaml": { "parsable": 625, "notParsable": 17, "empty": 2582 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 400, "counter": 13 }, { "number": 0, "counter": 12 }, { "number": -18, "counter": 2 }, { "number": 13, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 75, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 440 }, { "number": -1, "counter": 177 }, { "number": 1, "counter": 74 }, { "number": 14, "counter": 65 }, { "number": 15, "counter": 50 }, { "number": 3, "counter": 48 }, { "number": 10, "counter": 48 }, { "number": 18, "counter": 47 }, { "number": 16, "counter": 43 }, { "number": 12, "counter": 40 }, { "number": 4, "counter": 37 }, { "number": 5, "counter": 37 }, { "number": 6, "counter": 37 }, { "number": 2, "counter": 36 }, { "number": 8, "counter": 34 }, { "number": 17, "counter": 34 }, { "number": 9, "counter": 31 }, { "number": -10, "counter": 30 }, { "number": 20, "counter": 28 }, { "number": 24, "counter": 28 }, { "number": 7, "counter": 24 }, { "number": 23, "counter": 24 }, { "number": -7, "counter": 24 }, { "number": -8, "counter": 23 }, { "number": 11, "counter": 22 }, { "number": 21, "counter": 22 }, { "number": 25, "counter": 22 }, { "number": -13, "counter": 22 }, { "number": 19, "counter": 20 }, { "number": 100, "counter": 18 }, { "number": 22, "counter": 16 }, { "number": -9, "counter": 14 }, { "number": -12, "counter": 14 }, { "number": 27, "counter": 11 }, { "number": 28, "counter": 10 }, { "number": 30, "counter": 10 }, { "number": -3, "counter": 10 }, { "number": -4, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": -11, "counter": 9 }, { "number": 33, "counter": 8 }, { "number": 37, "counter": 8 }, { "number": 224, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": -2, "counter": 8 }, { "number": 43, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 211, "counter": 6 } ], "valid": 506, "invalid": 136, "ratioPresent": 0.9053632766542818, "nbUnion": 35353, "duplicates": 300, "invented": 91 }, "meta": { "withName": 610, "withoutName": 32, "withMolfile": 464 }, "info": { "protonNMR": { "present": 584, "notPresent": 58, "valid": 229, "notValid": 355 }, "carbonNMR": { "present": 434, "notPresent": 208, "valid": 168, "notValid": 266 }, "elementalAnalysis": { "present": 258, "notPresent": 384, "valid": 218, "notValid": 40 }, "hrms": { "present": 79, "notPresent": 563, "valid": 14, "notValid": 65 } }, "logs": { "info": 66, "warn": 225, "error": 173, "fatal": 178 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 468941, "totalTokens": 945678, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.414619 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.3_0.3", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 625 ], "notParsable": [ 17 ], "empty": [ 2582 ] }, "numbers": { "valid": [ 506 ], "invalid": [ 136 ], "ratioPresent": [ 0.9053632766542818 ], "nbUnion": [ 35353 ], "duplicates": [ 300 ], "invented": [ 91 ] }, "meta": { "withName": [ 610 ], "withoutName": [ 32 ], "withMolfile": [ 464 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_0.3_0.3", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 569, "frequency": { "50": 4, "80": 2, "200": 4, "300": 9, "400": 40, "500": 3 }, "solvent": { "CDCl3": 91, "not specified": 43, "DMSO-d6": 71, "CD3OD": 33, "Not specified": 6, "unknown": 8, "CHCl3": 1, "CD3CN": 6, "MeOH": 9, "CD3COCD3": 1, "CDC13": 1, "pyridine-d5": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 443, "frequency": { "50": 2, "75": 5, "80": 2, "100": 36, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1, "125.7": 1 }, "solvent": { "DMSO-d6": 50, "not specified": 31, "CDCl3": 76, "CD3OD": 31, "unknown": 5, "CHCl3": 2, "MeOH": 6, "CD3COCD3": 1, "Not specified": 5, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "CH3OH": 1 } } }, "yaml": { "parsable": 607, "notParsable": 35, "empty": 2222 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 46 }, { "number": 400, "counter": 22 }, { "number": 0, "counter": 10 }, { "number": 50, "counter": 6 }, { "number": 29, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 11, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 296.5, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": -23.86, "counter": 1 }, { "number": 9.71, "counter": 1 }, { "number": 9.8, "counter": 1 }, { "number": 21.67, "counter": 1 }, { "number": 58.82, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 125.7, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 458.2717, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 431 }, { "number": -1, "counter": 186 }, { "number": 1, "counter": 133 }, { "number": 3, "counter": 70 }, { "number": 2, "counter": 61 }, { "number": 6, "counter": 54 }, { "number": 8, "counter": 52 }, { "number": 5, "counter": 50 }, { "number": 4, "counter": 49 }, { "number": 10, "counter": 40 }, { "number": 14, "counter": 39 }, { "number": 12, "counter": 32 }, { "number": 100, "counter": 32 }, { "number": 7, "counter": 30 }, { "number": 9, "counter": 30 }, { "number": 15, "counter": 29 }, { "number": 16, "counter": 27 }, { "number": 11, "counter": 24 }, { "number": 17, "counter": 24 }, { "number": -3, "counter": 24 }, { "number": 18, "counter": 23 }, { "number": 25, "counter": 23 }, { "number": 24, "counter": 21 }, { "number": -2, "counter": 21 }, { "number": -4, "counter": 20 }, { "number": 20, "counter": 17 }, { "number": 23, "counter": 16 }, { "number": 28, "counter": 16 }, { "number": 19, "counter": 14 }, { "number": 21, "counter": 13 }, { "number": 224, "counter": 13 }, { "number": -5, "counter": 13 }, { "number": 211, "counter": 12 }, { "number": 70, "counter": 11 }, { "number": 50, "counter": 10 }, { "number": 196, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": 22, "counter": 9 }, { "number": 63, "counter": 9 }, { "number": 254, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": -6, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 90, "counter": 7 }, { "number": 282, "counter": 7 }, { "number": 360, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 27, "counter": 6 } ], "valid": 472, "invalid": 170, "ratioPresent": 0.882222987247776, "nbUnion": 35408, "duplicates": 361, "invented": 150 }, "meta": { "withName": 589, "withoutName": 53, "withMolfile": 452 }, "info": { "protonNMR": { "present": 570, "notPresent": 72, "valid": 200, "notValid": 370 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 158, "notValid": 285 }, "elementalAnalysis": { "present": 236, "notPresent": 406, "valid": 195, "notValid": 41 }, "hrms": { "present": 108, "notPresent": 534, "valid": 14, "notValid": 94 } }, "logs": { "info": 58, "warn": 151, "error": 243, "fatal": 190 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 466912, "totalTokens": 941081, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.407993 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_0.3_0.3", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 607 ], "notParsable": [ 35 ], "empty": [ 2222 ] }, "numbers": { "valid": [ 472 ], "invalid": [ 170 ], "ratioPresent": [ 0.882222987247776 ], "nbUnion": [ 35408 ], "duplicates": [ 361 ], "invented": [ 150 ] }, "meta": { "withName": [ 589 ], "withoutName": [ 53 ], "withMolfile": [ 452 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_0.3_0.3", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.3-0.3_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.3-0.3_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 586, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 5, "16.9": 1 }, "solvent": { "CDCl3": 84, "Not specified": 1, "DMSO-d6": 74, "CD3OD": 31, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 9, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "MeOD": 3, "qz": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 429, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1, "16.9": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 30, "CHCl3": 2, "CDCl3": 66, "CD3COCD3": 1, "N/A": 2, "unknown": 1, "pyridine-d5": 6, "MeOH": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "ppm": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 1700 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 13 }, { "number": 100, "counter": 11 }, { "number": 400, "counter": 6 }, { "number": 29, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 22, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 436 }, { "number": -1, "counter": 177 }, { "number": 1, "counter": 81 }, { "number": 14, "counter": 53 }, { "number": 10, "counter": 47 }, { "number": 4, "counter": 46 }, { "number": 16, "counter": 45 }, { "number": 3, "counter": 43 }, { "number": 15, "counter": 42 }, { "number": 18, "counter": 41 }, { "number": 6, "counter": 39 }, { "number": 9, "counter": 36 }, { "number": 12, "counter": 36 }, { "number": 8, "counter": 35 }, { "number": 5, "counter": 33 }, { "number": 17, "counter": 33 }, { "number": 2, "counter": 28 }, { "number": 24, "counter": 27 }, { "number": 25, "counter": 24 }, { "number": -10, "counter": 24 }, { "number": 7, "counter": 23 }, { "number": 19, "counter": 23 }, { "number": 23, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 21, "counter": 19 }, { "number": 100, "counter": 19 }, { "number": -9, "counter": 17 }, { "number": -7, "counter": 17 }, { "number": 20, "counter": 16 }, { "number": 22, "counter": 16 }, { "number": -8, "counter": 16 }, { "number": 28, "counter": 15 }, { "number": 27, "counter": 11 }, { "number": 30, "counter": 11 }, { "number": -2, "counter": 11 }, { "number": -13, "counter": 11 }, { "number": -3, "counter": 11 }, { "number": 211, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 196, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": -12, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": 225, "counter": 6 }, { "number": 240, "counter": 6 } ], "valid": 530, "invalid": 112, "ratioPresent": 0.9094466023307278, "nbUnion": 35339, "duplicates": 270, "invented": 82 }, "meta": { "withName": 614, "withoutName": 28, "withMolfile": 462 }, "info": { "protonNMR": { "present": 587, "notPresent": 55, "valid": 229, "notValid": 358 }, "carbonNMR": { "present": 430, "notPresent": 212, "valid": 171, "notValid": 259 }, "elementalAnalysis": { "present": 251, "notPresent": 391, "valid": 214, "notValid": 37 }, "hrms": { "present": 91, "notPresent": 551, "valid": 10, "notValid": 81 } }, "logs": { "info": 68, "warn": 204, "error": 190, "fatal": 180 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 441283, "totalTokens": 923798, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 0.3, "top_p": 0.3, "cost": 1.365081 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_0.3_0.3", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 639 ], "notParsable": [ 3 ], "empty": [ 1700 ] }, "numbers": { "valid": [ 530 ], "invalid": [ 112 ], "ratioPresent": [ 0.9094466023307278 ], "nbUnion": [ 35339 ], "duplicates": [ 270 ], "invented": [ 82 ] }, "meta": { "withName": [ 614 ], "withoutName": [ 28 ], "withMolfile": [ 462 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 29, "500": 6, "Not specified": 3, "not specified": 1 }, "solvent": { "CDCl3": 102, "not specified": 14, "not provided": 1, "CD3OD": 32, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "Not specified": 5, "N/A": 2, "unknown": 4, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "unspecified": 3, "Not available": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 431, "frequency": { "50": 2, "75": 5, "80": 2, "100": 25, "125": 3, "200": 2, "Not specified": 3, "100.6": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 29, "CDCl3": 82, "null": 1, "CD3COCD3": 1, "Not specified": 5, "N/A": 2, "not specified": 10, "pyridine-d5": 7, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "unknown": 3, "Not available": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "unspecified": 1 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2186 }, "numbers": { "inventedNumbers": [ { "number": 400, "counter": 11 }, { "number": 100, "counter": 10 }, { "number": 15, "counter": 3 }, { "number": 56, "counter": 3 }, { "number": 22, "counter": 2 }, { "number": 27, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": 43, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": -3, "counter": 2 }, { "number": 0, "counter": 1 }, { "number": 7, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 4.05, "counter": 1 }, { "number": 23.49, "counter": 1 }, { "number": 72.46, "counter": 1 }, { "number": 3.83, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 432 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 100 }, { "number": 14, "counter": 55 }, { "number": 10, "counter": 40 }, { "number": 16, "counter": 39 }, { "number": 8, "counter": 38 }, { "number": 18, "counter": 38 }, { "number": 5, "counter": 37 }, { "number": 9, "counter": 37 }, { "number": 4, "counter": 36 }, { "number": 12, "counter": 34 }, { "number": 15, "counter": 34 }, { "number": 6, "counter": 31 }, { "number": 2, "counter": 30 }, { "number": 3, "counter": 29 }, { "number": 17, "counter": 29 }, { "number": 19, "counter": 25 }, { "number": 24, "counter": 25 }, { "number": 25, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 11, "counter": 20 }, { "number": 20, "counter": 20 }, { "number": 21, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 28, "counter": 15 }, { "number": 100, "counter": 13 }, { "number": 224, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 27, "counter": 10 }, { "number": 22, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 70, "counter": 8 }, { "number": 225, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": 196, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 36, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 43, "counter": 5 }, { "number": 48, "counter": 5 }, { "number": 83, "counter": 5 } ], "valid": 534, "invalid": 108, "ratioPresent": 0.9179281410629384, "nbUnion": 35382, "duplicates": 297, "invented": 104 }, "meta": { "withName": 637, "withoutName": 5, "withMolfile": 481 }, "info": { "protonNMR": { "present": 596, "notPresent": 46, "valid": 226, "notValid": 370 }, "carbonNMR": { "present": 431, "notPresent": 211, "valid": 165, "notValid": 266 }, "elementalAnalysis": { "present": 267, "notPresent": 375, "valid": 216, "notValid": 51 }, "hrms": { "present": 80, "notPresent": 562, "valid": 14, "notValid": 66 } }, "logs": { "info": 75, "warn": 207, "error": 199, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450420, "totalTokens": 970171, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.420591 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v1" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v101.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v101" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 1" }, "stats": { "nmr": { "proton": { "nbSpectra": 592, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 6 }, "solvent": { "CDCl3": 94, "uncertain": 1, "not specified": 18, "CD3OD": 34, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unknown": 5, "unspecified": 1, "CDC13": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 4, "Not specified": 2, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 438, "frequency": { "50": 2, "75": 5, "80": 2, "100": 23, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CD3OD": 31, "CDCl3": 76, "unknown": 3, "unspecified": 2, "CD3COCD3": 1, "N/A": 3, "pyridine-d5": 7, "not specified": 11, "MeOH": 3, "Not specified": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2121 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 18 }, { "number": 400, "counter": 6 }, { "number": 15, "counter": 4 }, { "number": 29, "counter": 3 }, { "number": 30, "counter": 3 }, { "number": 14, "counter": 2 }, { "number": 21, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 0, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 7, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 38, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 123, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 6.06, "counter": 1 }, { "number": 7.12, "counter": 1 }, { "number": 30.05, "counter": 1 }, { "number": 56.77, "counter": 1 }, { "number": 5.94, "counter": 1 }, { "number": 29.54, "counter": 1 }, { "number": 64.52, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 159 }, { "number": 1, "counter": 92 }, { "number": 14, "counter": 50 }, { "number": 10, "counter": 42 }, { "number": 8, "counter": 38 }, { "number": 15, "counter": 38 }, { "number": 16, "counter": 37 }, { "number": 9, "counter": 35 }, { "number": 5, "counter": 34 }, { "number": 3, "counter": 33 }, { "number": 18, "counter": 32 }, { "number": 4, "counter": 31 }, { "number": 12, "counter": 30 }, { "number": 17, "counter": 30 }, { "number": 6, "counter": 29 }, { "number": 2, "counter": 27 }, { "number": 7, "counter": 21 }, { "number": 19, "counter": 21 }, { "number": 24, "counter": 21 }, { "number": 25, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 20, "counter": 18 }, { "number": 23, "counter": 18 }, { "number": 21, "counter": 14 }, { "number": 22, "counter": 12 }, { "number": 28, "counter": 12 }, { "number": 224, "counter": 9 }, { "number": -9, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 27, "counter": 7 }, { "number": 100, "counter": 7 }, { "number": 211, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": -14, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 30, "counter": 5 }, { "number": 33, "counter": 5 }, { "number": 34, "counter": 5 }, { "number": 70, "counter": 5 }, { "number": 196, "counter": 5 }, { "number": 210, "counter": 5 }, { "number": 239, "counter": 5 } ], "valid": 526, "invalid": 116, "ratioPresent": 0.9212763048069, "nbUnion": 35379, "duplicates": 256, "invented": 95 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 481 }, "info": { "protonNMR": { "present": 592, "notPresent": 50, "valid": 232, "notValid": 360 }, "carbonNMR": { "present": 438, "notPresent": 204, "valid": 171, "notValid": 267 }, "elementalAnalysis": { "present": 262, "notPresent": 380, "valid": 219, "notValid": 43 }, "hrms": { "present": 82, "notPresent": 560, "valid": 14, "notValid": 68 } }, "logs": { "info": 76, "warn": 204, "error": 201, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450973, "totalTokens": 970724, "nbPrompts": 642, "version": "v101", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.421697 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v101" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v102.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v102" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 2" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 5, "Not specified": 2, "not specified": 1 }, "solvent": { "CDCl3": 95, "not specified": 13, "uncertain": 1, "CD3OD": 34, "DMSO-d6": 71, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "unspecified": 2, "unknown": 3, "CDC13": 1, "Not specified": 5, "N/A": 2, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 434, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "Not specified": 1, "100.6": 1, "50.3": 1, "75.4": 1, "not specified": 1 }, "solvent": { "DMSO-d6": 50, "uncertain": 1, "not specified": 8, "CD3OD": 31, "CDCl3": 78, "null": 2, "unspecified": 1, "unknown": 2, "CD3COCD3": 1, "Not specified": 4, "N/A": 2, "pyridine-d5": 7, "MeOH": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2209 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 400, "counter": 6 }, { "number": 50, "counter": 5 }, { "number": 15, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 24, "counter": 2 }, { "number": 38, "counter": 2 }, { "number": 56, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 13, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 16, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 54, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.89, "counter": 1 }, { "number": 10.75, "counter": 1 }, { "number": 20.84, "counter": 1 }, { "number": 64.52, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 5.85, "counter": 1 }, { "number": 5.93, "counter": 1 }, { "number": 7.56, "counter": 1 }, { "number": 7.63, "counter": 1 }, { "number": 26.36, "counter": 1 }, { "number": 26.41, "counter": 1 }, { "number": 60.11, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 169 }, { "number": 1, "counter": 94 }, { "number": 14, "counter": 46 }, { "number": 10, "counter": 43 }, { "number": 16, "counter": 42 }, { "number": 8, "counter": 39 }, { "number": 15, "counter": 39 }, { "number": 18, "counter": 36 }, { "number": 2, "counter": 35 }, { "number": 5, "counter": 35 }, { "number": 6, "counter": 34 }, { "number": 12, "counter": 34 }, { "number": 3, "counter": 32 }, { "number": 4, "counter": 31 }, { "number": 9, "counter": 31 }, { "number": 17, "counter": 27 }, { "number": 24, "counter": 27 }, { "number": 7, "counter": 23 }, { "number": 19, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 20, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 25, "counter": 18 }, { "number": 21, "counter": 17 }, { "number": -7, "counter": 15 }, { "number": 28, "counter": 14 }, { "number": -8, "counter": 14 }, { "number": 100, "counter": 12 }, { "number": -10, "counter": 12 }, { "number": -6, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": -4, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": -5, "counter": 9 }, { "number": -2, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -13, "counter": 8 }, { "number": 27, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": -3, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 36, "counter": 6 }, { "number": 196, "counter": 6 } ], "valid": 518, "invalid": 124, "ratioPresent": 0.9166519655720712, "nbUnion": 35387, "duplicates": 222, "invented": 106 }, "meta": { "withName": 635, "withoutName": 7, "withMolfile": 479 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 232, "notValid": 365 }, "carbonNMR": { "present": 435, "notPresent": 207, "valid": 171, "notValid": 264 }, "elementalAnalysis": { "present": 263, "notPresent": 379, "valid": 216, "notValid": 47 }, "hrms": { "present": 90, "notPresent": 552, "valid": 14, "notValid": 76 } }, "logs": { "info": 74, "warn": 192, "error": 213, "fatal": 163 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 449854, "totalTokens": 969605, "nbPrompts": 642, "version": "v102", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.4194589999999998 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v102" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v103.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v103" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 3" }, "stats": { "nmr": { "proton": { "nbSpectra": 592, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 6 }, "solvent": { "CDCl3": 94, "not specified": 10, "CD3OD": 32, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "unspecified": 1, "CDC13": 1, "N/A": 1, "Not specified": 2, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "acetone-d6": 3, "unknown": 1 } }, "carbon": { "nbSpectra": 437, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "100.6": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 30, "null": 1, "unspecified": 1, "CD3COCD3": 1, "N/A": 1, "Not specified": 2, "pyridine-d5": 7, "MeOH": 5, "not specified": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "unknown": 1 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 2145 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 16 }, { "number": 400, "counter": 4 }, { "number": 15, "counter": 3 }, { "number": 39, "counter": 2 }, { "number": 3.83, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2.71, "counter": 2 }, { "number": 0, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 7.47, "counter": 1 }, { "number": 7.54, "counter": 1 }, { "number": 28.27, "counter": 1 }, { "number": 28.45, "counter": 1 }, { "number": 60.21, "counter": 1 }, { "number": 60.43, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 2.98, "counter": 1 }, { "number": 8.37, "counter": 1 }, { "number": 79.84, "counter": 1 }, { "number": 54.06, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 159 }, { "number": 1, "counter": 97 }, { "number": 6, "counter": 42 }, { "number": 10, "counter": 42 }, { "number": 8, "counter": 41 }, { "number": 14, "counter": 41 }, { "number": 16, "counter": 36 }, { "number": 4, "counter": 35 }, { "number": 12, "counter": 35 }, { "number": 3, "counter": 33 }, { "number": 18, "counter": 33 }, { "number": 15, "counter": 32 }, { "number": 5, "counter": 31 }, { "number": 2, "counter": 30 }, { "number": 9, "counter": 30 }, { "number": 17, "counter": 30 }, { "number": 7, "counter": 23 }, { "number": 25, "counter": 21 }, { "number": 19, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 23, "counter": 18 }, { "number": 24, "counter": 18 }, { "number": 28, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 20, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": 27, "counter": 11 }, { "number": 211, "counter": 11 }, { "number": 224, "counter": 11 }, { "number": 239, "counter": 11 }, { "number": 196, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": 282, "counter": 8 }, { "number": -10, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": 26, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 5 }, { "number": 30, "counter": 5 }, { "number": 36, "counter": 5 } ], "valid": 540, "invalid": 102, "ratioPresent": 0.9177548778393566, "nbUnion": 35362, "duplicates": 204, "invented": 89 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 483 }, "info": { "protonNMR": { "present": 594, "notPresent": 48, "valid": 226, "notValid": 368 }, "carbonNMR": { "present": 440, "notPresent": 202, "valid": 170, "notValid": 270 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 213, "notValid": 48 }, "hrms": { "present": 78, "notPresent": 564, "valid": 13, "notValid": 65 } }, "logs": { "info": 75, "warn": 206, "error": 202, "fatal": 159 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 447778, "totalTokens": 967529, "nbPrompts": 642, "version": "v103", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.4153069999999999 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v103" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v104.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v104" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 4" }, "stats": { "nmr": { "proton": { "nbSpectra": 592, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 6 }, "solvent": { "CDCl3": 94, "not specified": 12, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 2, "CDC13": 1, "Not specified": 3, "N/A": 1, "pyridine-d5": 7, "MeOH": 6, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "unknown": 1, "acetone-d6": 3, "unspecified": 1 } }, "carbon": { "nbSpectra": 433, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 77, "CD3OD": 30, "null": 2, "CD3COCD3": 1, "Not specified": 3, "N/A": 1, "pyridine-d5": 7, "MeOH": 6, "not specified": 8, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "unspecified": 1 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2043 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 0, "counter": 5 }, { "number": 400, "counter": 3 }, { "number": 15, "counter": 2 }, { "number": 28, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 7, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.88, "counter": 1 }, { "number": 5.34, "counter": 1 }, { "number": 26.15, "counter": 1 }, { "number": 64.63, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 3.22, "counter": 1 }, { "number": 7.54, "counter": 1 }, { "number": 7.71, "counter": 1 }, { "number": 28.37, "counter": 1 }, { "number": 28.64, "counter": 1 }, { "number": 60.65, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 427 }, { "number": -1, "counter": 166 }, { "number": 1, "counter": 85 }, { "number": 14, "counter": 45 }, { "number": 10, "counter": 41 }, { "number": 15, "counter": 36 }, { "number": 16, "counter": 35 }, { "number": 8, "counter": 34 }, { "number": 9, "counter": 32 }, { "number": 18, "counter": 31 }, { "number": 12, "counter": 30 }, { "number": 5, "counter": 29 }, { "number": 6, "counter": 28 }, { "number": 2, "counter": 27 }, { "number": 3, "counter": 27 }, { "number": 4, "counter": 27 }, { "number": 17, "counter": 25 }, { "number": 24, "counter": 23 }, { "number": 19, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 11, "counter": 17 }, { "number": 20, "counter": 17 }, { "number": 25, "counter": 17 }, { "number": 7, "counter": 16 }, { "number": 28, "counter": 14 }, { "number": 21, "counter": 13 }, { "number": 22, "counter": 12 }, { "number": -10, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": -8, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 27, "counter": 9 }, { "number": 100, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": 282, "counter": 9 }, { "number": -7, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 32, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 239, "counter": 7 }, { "number": -13, "counter": 7 }, { "number": -12, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 70, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": 267, "counter": 6 }, { "number": -14, "counter": 6 }, { "number": -5, "counter": 6 } ], "valid": 534, "invalid": 108, "ratioPresent": 0.9231612579176597, "nbUnion": 35392, "duplicates": 209, "invented": 120 }, "meta": { "withName": 639, "withoutName": 3, "withMolfile": 482 }, "info": { "protonNMR": { "present": 592, "notPresent": 50, "valid": 230, "notValid": 362 }, "carbonNMR": { "present": 433, "notPresent": 209, "valid": 171, "notValid": 262 }, "elementalAnalysis": { "present": 267, "notPresent": 375, "valid": 216, "notValid": 51 }, "hrms": { "present": 82, "notPresent": 560, "valid": 11, "notValid": 71 } }, "logs": { "info": 72, "warn": 200, "error": 210, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448708, "totalTokens": 968459, "nbPrompts": 642, "version": "v104", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.417167 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v104" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v105.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v105" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 592, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 4 }, "solvent": { "CDCl3": 96, "not specified": 7, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "CD3OD": 34, "unspecified": 1, "null": 1, "CDC13": 1, "N/A": 4, "pyridine-d5": 7, "MeOH": 5, "Not specified": 3, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "unknown": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 441, "frequency": { "50": 2, "75": 5, "80": 2, "100": 23, "125": 2, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 78, "CD3OD": 31, "unspecified": 1, "null": 1, "CD3COCD3": 1, "N/A": 4, "pyridine-d5": 7, "MeOH": 5, "not specified": 4, "Not specified": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 2109 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 0, "counter": 5 }, { "number": 400, "counter": 5 }, { "number": 15, "counter": 3 }, { "number": 50, "counter": 3 }, { "number": 29, "counter": 2 }, { "number": 3.83, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 4, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 291, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.57, "counter": 1 }, { "number": 10.73, "counter": 1 }, { "number": 11.43, "counter": 1 }, { "number": 20.71, "counter": 1 }, { "number": 21.05, "counter": 1 }, { "number": 64.29, "counter": 1 }, { "number": 64.39, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 5.94, "counter": 1 }, { "number": 29.31, "counter": 1 }, { "number": 64.75, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 4.08, "counter": 1 }, { "number": 22.99, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 426 }, { "number": -1, "counter": 165 }, { "number": 1, "counter": 94 }, { "number": 14, "counter": 47 }, { "number": 10, "counter": 42 }, { "number": 16, "counter": 42 }, { "number": 6, "counter": 37 }, { "number": 15, "counter": 37 }, { "number": 8, "counter": 35 }, { "number": 2, "counter": 34 }, { "number": 18, "counter": 33 }, { "number": 9, "counter": 32 }, { "number": 12, "counter": 32 }, { "number": 5, "counter": 31 }, { "number": 3, "counter": 30 }, { "number": 4, "counter": 30 }, { "number": 24, "counter": 26 }, { "number": 17, "counter": 23 }, { "number": 7, "counter": 22 }, { "number": 20, "counter": 21 }, { "number": 11, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 25, "counter": 18 }, { "number": 21, "counter": 17 }, { "number": 28, "counter": 15 }, { "number": 22, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": -7, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": -10, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 32, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -9, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": -5, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": -3, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": -2, "counter": 6 }, { "number": -13, "counter": 6 } ], "valid": 531, "invalid": 111, "ratioPresent": 0.9193575980137831, "nbUnion": 35375, "duplicates": 286, "invented": 100 }, "meta": { "withName": 636, "withoutName": 6, "withMolfile": 481 }, "info": { "protonNMR": { "present": 593, "notPresent": 49, "valid": 235, "notValid": 358 }, "carbonNMR": { "present": 443, "notPresent": 199, "valid": 172, "notValid": 271 }, "elementalAnalysis": { "present": 262, "notPresent": 380, "valid": 215, "notValid": 47 }, "hrms": { "present": 82, "notPresent": 560, "valid": 12, "notValid": 70 } }, "logs": { "info": 75, "warn": 205, "error": 201, "fatal": 161 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450371, "totalTokens": 970122, "nbPrompts": 642, "version": "v105", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.420493 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v105" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v106.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v106" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 5" }, "stats": { "nmr": { "proton": { "nbSpectra": 596, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 22, "500": 6, "not specified": 1 }, "solvent": { "CDCl3": 95, "not specified": 5, "not provided": 1, "uncertain": 1, "CD3OD": 32, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "unknown": 2, "CDC13": 1, "N/A": 3, "Not specified": 5, "pyridine-d5": 7, "MeOH": 5, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 438, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CD3OD": 29, "CDCl3": 77, "unknown": 1, "CD3COCD3": 1, "N/A": 3, "Not specified": 5, "pyridine-d5": 7, "MeOH": 3, "not specified": 3, "TBD": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 2172 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 23 }, { "number": 0, "counter": 4 }, { "number": 400, "counter": 4 }, { "number": 15, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 33, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 3.39, "counter": 2 }, { "number": 7, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 58, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 5.47, "counter": 1 }, { "number": 5.58, "counter": 1 }, { "number": 19.09, "counter": 1 }, { "number": 19.15, "counter": 1 }, { "number": 63.27, "counter": 1 }, { "number": 63.41, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 431 }, { "number": -1, "counter": 169 }, { "number": 1, "counter": 92 }, { "number": 14, "counter": 45 }, { "number": 10, "counter": 41 }, { "number": 15, "counter": 41 }, { "number": 16, "counter": 41 }, { "number": 8, "counter": 38 }, { "number": 18, "counter": 36 }, { "number": 12, "counter": 35 }, { "number": 4, "counter": 34 }, { "number": 6, "counter": 34 }, { "number": 9, "counter": 32 }, { "number": 5, "counter": 30 }, { "number": 3, "counter": 29 }, { "number": 2, "counter": 26 }, { "number": 17, "counter": 26 }, { "number": 24, "counter": 24 }, { "number": 7, "counter": 21 }, { "number": 11, "counter": 20 }, { "number": 19, "counter": 20 }, { "number": 23, "counter": 18 }, { "number": 20, "counter": 17 }, { "number": 21, "counter": 16 }, { "number": 25, "counter": 16 }, { "number": 28, "counter": 14 }, { "number": -7, "counter": 14 }, { "number": -10, "counter": 13 }, { "number": 22, "counter": 12 }, { "number": 100, "counter": 12 }, { "number": -9, "counter": 12 }, { "number": 224, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 211, "counter": 10 }, { "number": 239, "counter": 9 }, { "number": -8, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -13, "counter": 7 }, { "number": -12, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": 29, "counter": 5 } ], "valid": 529, "invalid": 113, "ratioPresent": 0.9212416853933607, "nbUnion": 35384, "duplicates": 243, "invented": 107 }, "meta": { "withName": 638, "withoutName": 4, "withMolfile": 482 }, "info": { "protonNMR": { "present": 596, "notPresent": 46, "valid": 232, "notValid": 364 }, "carbonNMR": { "present": 438, "notPresent": 204, "valid": 173, "notValid": 265 }, "elementalAnalysis": { "present": 266, "notPresent": 376, "valid": 219, "notValid": 47 }, "hrms": { "present": 88, "notPresent": 554, "valid": 13, "notValid": 75 } }, "logs": { "info": 78, "warn": 193, "error": 211, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 450295, "totalTokens": 970046, "nbPrompts": 642, "version": "v106", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.420341 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v106" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v107.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v107" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 7" }, "stats": { "nmr": { "proton": { "nbSpectra": 599, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 6, "not specified": 1 }, "solvent": { "CDCl3": 94, "not specified": 11, "CD3OD": 35, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "Not specified": 2, "unspecified": 1, "null": 1, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 5, "unknown": 1, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "not available": 1, "acetone-d6": 3, "uncertain": 2 } }, "carbon": { "nbSpectra": 447, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 77, "CD3OD": 32, "Not specified": 2, "unspecified": 1, "null": 1, "CD3COCD3": 1, "N/A": 2, "not specified": 8, "pyridine-d5": 6, "MeOH": 5, "unknown": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "uncertain": 2 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 2283 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 20 }, { "number": 15, "counter": 4 }, { "number": 0, "counter": 3 }, { "number": 29, "counter": 3 }, { "number": 50, "counter": 3 }, { "number": 400, "counter": 3 }, { "number": 21, "counter": 2 }, { "number": 28, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 35, "counter": 2 }, { "number": 39, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 12, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 22, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 6.31, "counter": 1 }, { "number": 6.42, "counter": 1 }, { "number": 19.78, "counter": 1 }, { "number": 19.91, "counter": 1 }, { "number": 60.65, "counter": 1 }, { "number": 60.74, "counter": 1 }, { "number": 5.03, "counter": 1 }, { "number": 5.15, "counter": 1 }, { "number": 17.12, "counter": 1 }, { "number": 17.39, "counter": 1 }, { "number": 17.62, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 430 }, { "number": -1, "counter": 165 }, { "number": 1, "counter": 87 }, { "number": 14, "counter": 52 }, { "number": 10, "counter": 40 }, { "number": 16, "counter": 37 }, { "number": 9, "counter": 34 }, { "number": 15, "counter": 34 }, { "number": 8, "counter": 33 }, { "number": 5, "counter": 31 }, { "number": 6, "counter": 31 }, { "number": 12, "counter": 30 }, { "number": 17, "counter": 30 }, { "number": 18, "counter": 30 }, { "number": 4, "counter": 29 }, { "number": 3, "counter": 27 }, { "number": 2, "counter": 25 }, { "number": 24, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 20, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 21, "counter": 18 }, { "number": 25, "counter": 18 }, { "number": 22, "counter": 13 }, { "number": 28, "counter": 13 }, { "number": -10, "counter": 12 }, { "number": -9, "counter": 11 }, { "number": -7, "counter": 11 }, { "number": 27, "counter": 9 }, { "number": 100, "counter": 9 }, { "number": 239, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 32, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 211, "counter": 7 }, { "number": 224, "counter": 7 }, { "number": 26, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 34, "counter": 5 }, { "number": 225, "counter": 5 }, { "number": 251, "counter": 5 } ], "valid": 527, "invalid": 115, "ratioPresent": 0.9262304083586941, "nbUnion": 35394, "duplicates": 267, "invented": 115 }, "meta": { "withName": 641, "withoutName": 1, "withMolfile": 482 }, "info": { "protonNMR": { "present": 599, "notPresent": 43, "valid": 232, "notValid": 367 }, "carbonNMR": { "present": 449, "notPresent": 193, "valid": 174, "notValid": 275 }, "elementalAnalysis": { "present": 265, "notPresent": 377, "valid": 220, "notValid": 45 }, "hrms": { "present": 74, "notPresent": 568, "valid": 11, "notValid": 63 } }, "logs": { "info": 76, "warn": 203, "error": 203, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 452728, "totalTokens": 972479, "nbPrompts": 642, "version": "v107", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.425207 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v107" }, { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v108.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v108" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Our baseline prompt. Replicate 8" }, "stats": { "nmr": { "proton": { "nbSpectra": 600, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 4 }, "solvent": { "CDCl3": 97, "not specified": 11, "CD3OD": 33, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "unknown": 2, "null": 1, "CDC13": 1, "Not specified": 3, "pyridine-d5": 7, "MeOH": 8, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 444, "frequency": { "50": 2, "75": 5, "80": 2, "100": 20, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 30, "CDCl3": 77, "unknown": 1, "CD3COCD3": 1, "Not specified": 2, "pyridine-d5": 6, "MeOH": 7, "not specified": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "see Table 1": 1, "acetone-d6": 5 } } }, "yaml": { "parsable": 641, "notParsable": 1, "empty": 1892 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 17 }, { "number": 400, "counter": 5 }, { "number": 0, "counter": 4 }, { "number": 30, "counter": 3 }, { "number": 29, "counter": 2 }, { "number": 41, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 9, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.14, "counter": 1 }, { "number": 8.01, "counter": 1 }, { "number": 79.22, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 6.14, "counter": 1 }, { "number": 6.15, "counter": 1 }, { "number": 29.23, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 430 }, { "number": -1, "counter": 160 }, { "number": 1, "counter": 96 }, { "number": 14, "counter": 51 }, { "number": 10, "counter": 40 }, { "number": 16, "counter": 40 }, { "number": 15, "counter": 38 }, { "number": 8, "counter": 36 }, { "number": 4, "counter": 35 }, { "number": 18, "counter": 35 }, { "number": 5, "counter": 34 }, { "number": 12, "counter": 34 }, { "number": 3, "counter": 32 }, { "number": 9, "counter": 31 }, { "number": 6, "counter": 30 }, { "number": 2, "counter": 27 }, { "number": 17, "counter": 26 }, { "number": 24, "counter": 21 }, { "number": 19, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 20, "counter": 19 }, { "number": 11, "counter": 18 }, { "number": 25, "counter": 18 }, { "number": 21, "counter": 15 }, { "number": 28, "counter": 15 }, { "number": 100, "counter": 13 }, { "number": 224, "counter": 12 }, { "number": 27, "counter": 11 }, { "number": 211, "counter": 11 }, { "number": 239, "counter": 10 }, { "number": 22, "counter": 9 }, { "number": 26, "counter": 9 }, { "number": 196, "counter": 9 }, { "number": 267, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 282, "counter": 8 }, { "number": -9, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": -10, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 29, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": 30, "counter": 5 }, { "number": 31, "counter": 5 }, { "number": 33, "counter": 5 }, { "number": 225, "counter": 5 }, { "number": 338, "counter": 5 } ], "valid": 524, "invalid": 118, "ratioPresent": 0.9178645481240065, "nbUnion": 35374, "duplicates": 222, "invented": 94 }, "meta": { "withName": 640, "withoutName": 2, "withMolfile": 484 }, "info": { "protonNMR": { "present": 601, "notPresent": 41, "valid": 234, "notValid": 367 }, "carbonNMR": { "present": 444, "notPresent": 198, "valid": 175, "notValid": 269 }, "elementalAnalysis": { "present": 255, "notPresent": 387, "valid": 212, "notValid": 43 }, "hrms": { "present": 82, "notPresent": 560, "valid": 14, "notValid": 68 } }, "logs": { "info": 77, "warn": 202, "error": 205, "fatal": 158 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 445436, "totalTokens": 965187, "nbPrompts": 642, "version": "v108", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.410623 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_0.5", "version": "v108" } ], "versions": [ "v1", "v101", "v102", "v103", "v104", "v105", "v106", "v107", "v108" ], "stats": { "yaml": { "parsable": [ 641, 641, 639, 640, 642, 639, 641, 642, 641 ], "notParsable": [ 1, 1, 3, 2, 0, 3, 1, 0, 1 ], "empty": [ 2186, 2121, 2209, 2145, 2043, 2109, 2172, 2283, 1892 ] }, "numbers": { "valid": [ 534, 526, 518, 540, 534, 531, 529, 527, 524 ], "invalid": [ 108, 116, 124, 102, 108, 111, 113, 115, 118 ], "ratioPresent": [ 0.9179281410629384, 0.9212763048069, 0.9166519655720712, 0.9177548778393566, 0.9231612579176597, 0.9193575980137831, 0.9212416853933607, 0.9262304083586941, 0.9178645481240065 ], "nbUnion": [ 35382, 35379, 35387, 35362, 35392, 35375, 35384, 35394, 35374 ], "duplicates": [ 297, 256, 222, 204, 209, 286, 243, 267, 222 ], "invented": [ 104, 95, 106, 89, 120, 100, 107, 115, 94 ] }, "meta": { "withName": [ 637, 639, 635, 639, 639, 636, 638, 641, 640 ], "withoutName": [ 5, 3, 7, 3, 3, 6, 4, 1, 2 ], "withMolfile": [ 481, 481, 479, 483, 482, 481, 482, 482, 484 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 568, "frequency": { "50": 4, "80": 2, "200": 3, "300": 11, "400": 43, "500": 7, "5.16": 1 }, "solvent": { "unknown": 94, "CDCl3": 94, "not specified": 21, "DMSO-d6": 73, "CD3OD": 31, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "N/A": 2, "literature": 1, "pyridine-d5": 7, "Not specified": 1, "MeOH": 6, "ppm": 3, "MeOD": 3, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "CHCl3-EtOH (1:1)": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 477, "frequency": { "50": 2, "75": 5, "80": 2, "100": 41, "125": 3, "200": 1, "400": 1, "50.3": 1, "75.4": 1, "178.8": 1 }, "solvent": { "unknown": 79, "DMSO-d6": 54, "not specified": 19, "CDCl3": 91, "CD3OD": 29, "CHCl3": 2, "CD3COCD3": 1, "N/A": 2, "literature": 1, "pyridine-d5": 6, "Not specified": 1, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "CHCl3-EtOH (1:1)": 1, "acetone-d6": 5 } } }, "yaml": { "parsable": 628, "notParsable": 14, "empty": 15593 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 154 }, { "number": 400, "counter": 28 }, { "number": 100, "counter": 11 }, { "number": -18, "counter": 2 }, { "number": 9, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 439 }, { "number": -1, "counter": 144 }, { "number": 1, "counter": 71 }, { "number": 14, "counter": 49 }, { "number": 4, "counter": 44 }, { "number": 15, "counter": 42 }, { "number": 5, "counter": 41 }, { "number": 100, "counter": 41 }, { "number": 8, "counter": 38 }, { "number": 18, "counter": 38 }, { "number": 3, "counter": 35 }, { "number": 16, "counter": 35 }, { "number": 10, "counter": 34 }, { "number": 6, "counter": 32 }, { "number": 12, "counter": 31 }, { "number": 17, "counter": 31 }, { "number": 9, "counter": 29 }, { "number": 2, "counter": 28 }, { "number": 11, "counter": 28 }, { "number": 24, "counter": 25 }, { "number": 7, "counter": 22 }, { "number": 19, "counter": 21 }, { "number": 20, "counter": 21 }, { "number": 25, "counter": 21 }, { "number": 21, "counter": 19 }, { "number": 211, "counter": 18 }, { "number": 224, "counter": 17 }, { "number": 239, "counter": 17 }, { "number": 23, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": -2, "counter": 16 }, { "number": 28, "counter": 15 }, { "number": 196, "counter": 15 }, { "number": 30, "counter": 14 }, { "number": -10, "counter": 14 }, { "number": -13, "counter": 13 }, { "number": -8, "counter": 13 }, { "number": 22, "counter": 12 }, { "number": 27, "counter": 12 }, { "number": -5, "counter": 12 }, { "number": -4, "counter": 12 }, { "number": 31, "counter": 11 }, { "number": 254, "counter": 11 }, { "number": 282, "counter": 11 }, { "number": -3, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 43, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": -6, "counter": 10 }, { "number": 32, "counter": 9 } ], "valid": 380, "invalid": 262, "ratioPresent": 0.894340003254072, "nbUnion": 35504, "duplicates": 317, "invented": 590 }, "meta": { "withName": 619, "withoutName": 23, "withMolfile": 472 }, "info": { "protonNMR": { "present": 589, "notPresent": 53, "valid": 163, "notValid": 426 }, "carbonNMR": { "present": 514, "notPresent": 128, "valid": 149, "notValid": 365 }, "elementalAnalysis": { "present": 274, "notPresent": 368, "valid": 229, "notValid": 45 }, "hrms": { "present": 55, "notPresent": 587, "valid": 9, "notValid": 46 } }, "logs": { "info": 38, "warn": 182, "error": 252, "fatal": 170 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 531136, "totalTokens": 1018145, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.549281 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.5_0.5", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 628 ], "notParsable": [ 14 ], "empty": [ 15593 ] }, "numbers": { "valid": [ 380 ], "invalid": [ 262 ], "ratioPresent": [ 0.894340003254072 ], "nbUnion": [ 35504 ], "duplicates": [ 317 ], "invented": [ 590 ] }, "meta": { "withName": [ 619 ], "withoutName": [ 23 ], "withMolfile": [ 472 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 583, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 26, "500": 4 }, "solvent": { "methanol": 1, "DMSO-d6": 72, "CDCl3": 87, "unknown": 10, "not specified": 4, "CD3OD": 32, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "N/A": 1, "pyridine-d5": 6, "MeOH": 5, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 3, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 434, "frequency": { "50": 2, "75": 5, "80": 2, "100": 25, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 49, "CDCl3": 72, "unknown": 7, "not specified": 3, "CD3OD": 31, "CHCl3": 1, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 6, "MeOH": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 625, "notParsable": 17, "empty": 2977 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 15 }, { "number": 0, "counter": 14 }, { "number": 400, "counter": 8 }, { "number": 29, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 58, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 438 }, { "number": -1, "counter": 173 }, { "number": 1, "counter": 79 }, { "number": 14, "counter": 59 }, { "number": 10, "counter": 50 }, { "number": 3, "counter": 48 }, { "number": 4, "counter": 45 }, { "number": 6, "counter": 45 }, { "number": 15, "counter": 44 }, { "number": 2, "counter": 42 }, { "number": 8, "counter": 40 }, { "number": 16, "counter": 39 }, { "number": 18, "counter": 37 }, { "number": 5, "counter": 35 }, { "number": 9, "counter": 34 }, { "number": 12, "counter": 34 }, { "number": 17, "counter": 34 }, { "number": 24, "counter": 27 }, { "number": 20, "counter": 26 }, { "number": 7, "counter": 24 }, { "number": 11, "counter": 24 }, { "number": 25, "counter": 24 }, { "number": 23, "counter": 23 }, { "number": -10, "counter": 22 }, { "number": 19, "counter": 21 }, { "number": -7, "counter": 20 }, { "number": 100, "counter": 19 }, { "number": 21, "counter": 18 }, { "number": -8, "counter": 17 }, { "number": 22, "counter": 16 }, { "number": -13, "counter": 15 }, { "number": 28, "counter": 13 }, { "number": 27, "counter": 12 }, { "number": 26, "counter": 10 }, { "number": -2, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": 30, "counter": 9 }, { "number": -12, "counter": 9 }, { "number": -6, "counter": 9 }, { "number": 31, "counter": 8 }, { "number": 43, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": -3, "counter": 8 }, { "number": 33, "counter": 7 }, { "number": 70, "counter": 6 }, { "number": 7.3, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 5 }, { "number": 37, "counter": 5 }, { "number": 225, "counter": 5 } ], "valid": 505, "invalid": 137, "ratioPresent": 0.8999398597690837, "nbUnion": 35346, "duplicates": 410, "invented": 88 }, "meta": { "withName": 604, "withoutName": 38, "withMolfile": 454 }, "info": { "protonNMR": { "present": 585, "notPresent": 57, "valid": 229, "notValid": 356 }, "carbonNMR": { "present": 435, "notPresent": 207, "valid": 170, "notValid": 265 }, "elementalAnalysis": { "present": 250, "notPresent": 392, "valid": 212, "notValid": 38 }, "hrms": { "present": 81, "notPresent": 561, "valid": 12, "notValid": 69 } }, "logs": { "info": 67, "warn": 216, "error": 171, "fatal": 188 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 466411, "totalTokens": 943148, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.409559 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.5_0.5", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 625 ], "notParsable": [ 17 ], "empty": [ 2977 ] }, "numbers": { "valid": [ 505 ], "invalid": [ 137 ], "ratioPresent": [ 0.8999398597690837 ], "nbUnion": [ 35346 ], "duplicates": [ 410 ], "invented": [ 88 ] }, "meta": { "withName": [ 604 ], "withoutName": [ 38 ], "withMolfile": [ 454 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_0.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 575, "frequency": { "50": 4, "80": 2, "200": 3, "300": 11, "400": 36, "500": 3, "Not specified": 1, "not specified": 1 }, "solvent": { "CDCl3": 88, "not specified": 34, "DMSO-d6": 74, "unknown": 7, "CD3OD": 32, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 10, "CDC13": 1, "Not specified": 9, "pyridine-d5": 7, "N/A": 2, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 442, "frequency": { "50": 2, "75": 5, "80": 2, "100": 30, "125": 3, "200": 1, "400": 2, "50.3": 1, "75.4": 1, "174.9": 1, "not specified": 1 }, "solvent": { "DMSO-d6": 52, "CDCl3": 77, "not specified": 26, "CD3OD": 30, "CHCl3": 1, "MeOH": 8, "unknown": 4, "CD3COCD3": 1, "Not specified": 7, "pyridine-d5": 7, "N/A": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "ppm": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 610, "notParsable": 32, "empty": 2014 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 35 }, { "number": 400, "counter": 20 }, { "number": 0, "counter": 9 }, { "number": 50, "counter": 8 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 47, "counter": 1 }, { "number": 75, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 10.43, "counter": 1 }, { "number": 78.16, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": -3, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": -18.7, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 2.89, "counter": 1 }, { "number": 8.36, "counter": 1 }, { "number": 10.92, "counter": 1 }, { "number": 77.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 431 }, { "number": -1, "counter": 183 }, { "number": 1, "counter": 124 }, { "number": 3, "counter": 69 }, { "number": 2, "counter": 54 }, { "number": 6, "counter": 52 }, { "number": 4, "counter": 49 }, { "number": 5, "counter": 48 }, { "number": 8, "counter": 47 }, { "number": 14, "counter": 41 }, { "number": 10, "counter": 35 }, { "number": 100, "counter": 34 }, { "number": 12, "counter": 32 }, { "number": 7, "counter": 30 }, { "number": 9, "counter": 30 }, { "number": 18, "counter": 29 }, { "number": 17, "counter": 27 }, { "number": 11, "counter": 25 }, { "number": 15, "counter": 25 }, { "number": 16, "counter": 23 }, { "number": 25, "counter": 21 }, { "number": -3, "counter": 21 }, { "number": 24, "counter": 20 }, { "number": -4, "counter": 18 }, { "number": 23, "counter": 17 }, { "number": -5, "counter": 16 }, { "number": 19, "counter": 14 }, { "number": 28, "counter": 14 }, { "number": -2, "counter": 14 }, { "number": 20, "counter": 13 }, { "number": 21, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": 63, "counter": 10 }, { "number": 211, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 50, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": 70, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": -10, "counter": 8 }, { "number": 27, "counter": 7 }, { "number": 90, "counter": 7 }, { "number": 91, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": 237, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": 8.6, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -11, "counter": 7 } ], "valid": 489, "invalid": 153, "ratioPresent": 0.887521379088022, "nbUnion": 35398, "duplicates": 305, "invented": 133 }, "meta": { "withName": 598, "withoutName": 44, "withMolfile": 457 }, "info": { "protonNMR": { "present": 576, "notPresent": 66, "valid": 212, "notValid": 364 }, "carbonNMR": { "present": 442, "notPresent": 200, "valid": 162, "notValid": 280 }, "elementalAnalysis": { "present": 247, "notPresent": 395, "valid": 207, "notValid": 40 }, "hrms": { "present": 108, "notPresent": 534, "valid": 14, "notValid": 94 } }, "logs": { "info": 67, "warn": 156, "error": 234, "fatal": 185 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 463021, "totalTokens": 937190, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.400211 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_0.5_0.5", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 610 ], "notParsable": [ 32 ], "empty": [ 2014 ] }, "numbers": { "valid": [ 489 ], "invalid": [ 153 ], "ratioPresent": [ 0.887521379088022 ], "nbUnion": [ 35398 ], "duplicates": [ 305 ], "invented": [ 133 ] }, "meta": { "withName": [ 598 ], "withoutName": [ 44 ], "withMolfile": [ 457 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_0.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-0.5_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-0.5_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 589, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 5 }, "solvent": { "CDCl3": 85, "not specified": 1, "DMSO-d6": 73, "CD3OD": 33, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 7, "MeOH": 9, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 432, "frequency": { "50": 2, "75": 5, "80": 2, "100": 17, "125": 3, "200": 2, "400": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 49, "CD3OD": 30, "CHCl3": 2, "CDCl3": 67, "CD3COCD3": 1, "pyridine-d5": 7, "MeOH": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 640, "notParsable": 2, "empty": 1685 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 12 }, { "number": 100, "counter": 9 }, { "number": 34, "counter": 3 }, { "number": 400, "counter": 3 }, { "number": -18, "counter": 2 }, { "number": 14, "counter": 1 }, { "number": 16, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 174 }, { "number": 1, "counter": 74 }, { "number": 14, "counter": 51 }, { "number": 4, "counter": 45 }, { "number": 18, "counter": 42 }, { "number": 10, "counter": 41 }, { "number": 16, "counter": 40 }, { "number": 3, "counter": 38 }, { "number": 15, "counter": 37 }, { "number": 6, "counter": 36 }, { "number": 8, "counter": 34 }, { "number": 17, "counter": 32 }, { "number": 5, "counter": 31 }, { "number": 9, "counter": 31 }, { "number": 2, "counter": 29 }, { "number": 12, "counter": 28 }, { "number": 24, "counter": 25 }, { "number": 25, "counter": 24 }, { "number": 19, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 20, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 23, "counter": 20 }, { "number": -10, "counter": 20 }, { "number": 21, "counter": 18 }, { "number": -9, "counter": 18 }, { "number": -7, "counter": 16 }, { "number": 100, "counter": 15 }, { "number": 28, "counter": 14 }, { "number": -8, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": 30, "counter": 10 }, { "number": -3, "counter": 10 }, { "number": -2, "counter": 10 }, { "number": -13, "counter": 9 }, { "number": -4, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 224, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 0.1, "counter": 6 }, { "number": 4.2, "counter": 6 }, { "number": -11, "counter": 6 }, { "number": -12, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": 29, "counter": 5 }, { "number": 34, "counter": 5 } ], "valid": 532, "invalid": 110, "ratioPresent": 0.9156517066339416, "nbUnion": 35335, "duplicates": 266, "invented": 69 }, "meta": { "withName": 618, "withoutName": 24, "withMolfile": 463 }, "info": { "protonNMR": { "present": 589, "notPresent": 53, "valid": 233, "notValid": 356 }, "carbonNMR": { "present": 432, "notPresent": 210, "valid": 171, "notValid": 261 }, "elementalAnalysis": { "present": 248, "notPresent": 394, "valid": 211, "notValid": 37 }, "hrms": { "present": 91, "notPresent": 551, "valid": 14, "notValid": 77 } }, "logs": { "info": 67, "warn": 213, "error": 183, "fatal": 179 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 440666, "totalTokens": 923181, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 0.5, "cost": 1.363847 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_0.5_0.5", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 640 ], "notParsable": [ 2 ], "empty": [ 1685 ] }, "numbers": { "valid": [ 532 ], "invalid": [ 110 ], "ratioPresent": [ 0.9156517066339416 ], "nbUnion": [ 35335 ], "duplicates": [ 266 ], "invented": [ 69 ] }, "meta": { "withName": [ 618 ], "withoutName": [ 24 ], "withMolfile": [ 463 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-1_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-1_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 583, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 21, "500": 5 }, "solvent": { "CDCl3": 80, "unknown": 4, "DMSO-d6": 74, "CD3OD": 33, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 1, "acetone-d6": 3, "not specified": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 426, "frequency": { "50": 2, "75": 5, "80": 2, "100": 18, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "CD3OD": 30, "CHCl3": 1, "CDCl3": 63, "CD3COCD3": 1, "in accordance with the literature [24]": 1, "pyridine-d5": 7, "MeOH": 5, "unknown": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "not specified": 1 } } }, "yaml": { "parsable": 636, "notParsable": 6, "empty": 1589 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 9 }, { "number": 0, "counter": 7 }, { "number": 50, "counter": 4 }, { "number": 400, "counter": 3 }, { "number": -18, "counter": 2 }, { "number": 12, "counter": 1 }, { "number": 16, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 436 }, { "number": -1, "counter": 183 }, { "number": 1, "counter": 97 }, { "number": 14, "counter": 54 }, { "number": 4, "counter": 47 }, { "number": 6, "counter": 44 }, { "number": 10, "counter": 43 }, { "number": 3, "counter": 38 }, { "number": 5, "counter": 35 }, { "number": 8, "counter": 35 }, { "number": 9, "counter": 35 }, { "number": 16, "counter": 35 }, { "number": 18, "counter": 35 }, { "number": 2, "counter": 33 }, { "number": 12, "counter": 33 }, { "number": 15, "counter": 33 }, { "number": -10, "counter": 30 }, { "number": 17, "counter": 27 }, { "number": -7, "counter": 27 }, { "number": 11, "counter": 25 }, { "number": 24, "counter": 23 }, { "number": -8, "counter": 23 }, { "number": 25, "counter": 22 }, { "number": 7, "counter": 21 }, { "number": 19, "counter": 21 }, { "number": 23, "counter": 20 }, { "number": -9, "counter": 19 }, { "number": -13, "counter": 18 }, { "number": 20, "counter": 16 }, { "number": 21, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": 100, "counter": 12 }, { "number": 28, "counter": 11 }, { "number": -2, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": -4, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 30, "counter": 9 }, { "number": -3, "counter": 9 }, { "number": -12, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": 224, "counter": 7 }, { "number": -14, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": 33, "counter": 6 }, { "number": 211, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 34, "counter": 5 } ], "valid": 553, "invalid": 89, "ratioPresent": 0.9113760487946652, "nbUnion": 35329, "duplicates": 198, "invented": 52 }, "meta": { "withName": 621, "withoutName": 21, "withMolfile": 471 }, "info": { "protonNMR": { "present": 583, "notPresent": 59, "valid": 204, "notValid": 379 }, "carbonNMR": { "present": 426, "notPresent": 216, "valid": 173, "notValid": 253 }, "elementalAnalysis": { "present": 252, "notPresent": 390, "valid": 210, "notValid": 42 }, "hrms": { "present": 86, "notPresent": 556, "valid": 12, "notValid": 74 } }, "logs": { "info": 68, "warn": 204, "error": 199, "fatal": 171 }, "chatgpt": { "promptTokens": 488935, "completionTokens": 438506, "totalTokens": 927441, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "cost": 1.365947 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_0.5_1", "version": "v1" } ], "versions": [ "v1" ], "stats": { "yaml": { "parsable": [ 636 ], "notParsable": [ 6 ], "empty": [ 1589 ] }, "numbers": { "valid": [ 553 ], "invalid": [ 89 ], "ratioPresent": [ 0.9113760487946652 ], "nbUnion": [ 35329 ], "duplicates": [ 198 ], "invented": [ 52 ] }, "meta": { "withName": [ 621 ], "withoutName": [ 21 ], "withMolfile": [ 471 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.5_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-1_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-1_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 569, "frequency": { "50": 4, "80": 2, "200": 3, "300": 12, "400": 61, "500": 7, "5.16": 1 }, "solvent": { "N/A": 6, "CDCl3": 101, "unknown": 83, "DMSO-d6": 73, "not specified": 18, "not provided": 1, "CD3OD": 30, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "literature": 1, "pyridine-d5": 6, "Not specified": 2, "MeOH": 8, "not available": 1, "MeOD": 3, "dimethylformamide": 1, "Unknown": 1, "D2O": 1, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "CHCl3-EtOH (1:1)": 2, "acetone-d6": 2 } }, "carbon": { "nbSpectra": 475, "frequency": { "50": 2, "75": 6, "80": 2, "100": 55, "125": 3, "200": 1, "50.3": 1, "75.4": 1, "178.8": 1 }, "solvent": { "DMSO-d6": 54, "unknown": 70, "not provided": 1, "CDCl3": 93, "not specified": 15, "CD3OD": 28, "CD3COCD3": 1, "N/A": 4, "literature": 1, "pyridine-d5": 6, "Not specified": 2, "MeOH": 5, "not available": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "ppm": 1, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "CHCl3-EtOH (1:1)": 1, "acetone-d6": 4 } } }, "yaml": { "parsable": 625, "notParsable": 17, "empty": 15162 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 172 }, { "number": 400, "counter": 46 }, { "number": 100, "counter": 21 }, { "number": 20, "counter": 2 }, { "number": 75, "counter": 2 }, { "number": 300, "counter": 2 }, { "number": 17, "counter": 1 }, { "number": 19, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": -18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 435 }, { "number": -1, "counter": 144 }, { "number": 1, "counter": 73 }, { "number": 14, "counter": 54 }, { "number": 4, "counter": 46 }, { "number": 5, "counter": 43 }, { "number": 100, "counter": 43 }, { "number": 16, "counter": 42 }, { "number": 15, "counter": 40 }, { "number": 18, "counter": 40 }, { "number": 17, "counter": 39 }, { "number": 8, "counter": 38 }, { "number": 3, "counter": 35 }, { "number": 2, "counter": 34 }, { "number": 10, "counter": 34 }, { "number": 12, "counter": 33 }, { "number": 9, "counter": 30 }, { "number": 6, "counter": 28 }, { "number": 11, "counter": 27 }, { "number": 7, "counter": 26 }, { "number": 24, "counter": 25 }, { "number": 19, "counter": 22 }, { "number": 20, "counter": 22 }, { "number": 25, "counter": 20 }, { "number": 21, "counter": 19 }, { "number": 224, "counter": 18 }, { "number": 22, "counter": 17 }, { "number": 23, "counter": 17 }, { "number": 28, "counter": 17 }, { "number": 211, "counter": 17 }, { "number": 239, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": 196, "counter": 14 }, { "number": -8, "counter": 14 }, { "number": 30, "counter": 13 }, { "number": -13, "counter": 12 }, { "number": -10, "counter": 12 }, { "number": -7, "counter": 12 }, { "number": 26, "counter": 11 }, { "number": 27, "counter": 11 }, { "number": 135, "counter": 11 }, { "number": 254, "counter": 11 }, { "number": -4, "counter": 11 }, { "number": -3, "counter": 11 }, { "number": -2, "counter": 11 }, { "number": 43, "counter": 10 }, { "number": 70, "counter": 10 }, { "number": 282, "counter": 10 }, { "number": -5, "counter": 10 }, { "number": 32, "counter": 9 } ], "valid": 353, "invalid": 289, "ratioPresent": 0.8906440272240513, "nbUnion": 35556, "duplicates": 371, "invented": 695 }, "meta": { "withName": 615, "withoutName": 27, "withMolfile": 470 }, "info": { "protonNMR": { "present": 588, "notPresent": 54, "valid": 169, "notValid": 419 }, "carbonNMR": { "present": 515, "notPresent": 127, "valid": 146, "notValid": 369 }, "elementalAnalysis": { "present": 270, "notPresent": 372, "valid": 225, "notValid": 45 }, "hrms": { "present": 57, "notPresent": 585, "valid": 7, "notValid": 50 } }, "logs": { "info": 33, "warn": 170, "error": 267, "fatal": 172 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 531272, "totalTokens": 1018281, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "cost": 1.5495530000000002 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_0.5_1", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 625 ], "notParsable": [ 17 ], "empty": [ 15162 ] }, "numbers": { "valid": [ 353 ], "invalid": [ 289 ], "ratioPresent": [ 0.8906440272240513 ], "nbUnion": [ 35556 ], "duplicates": [ 371 ], "invented": [ 695 ] }, "meta": { "withName": [ 615 ], "withoutName": [ 27 ], "withMolfile": [ 470 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.5_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-1_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-1_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 575, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 30, "500": 6 }, "solvent": { "CDCl3": 85, "not specified": 4, "N/A": 5, "DMSO-d6": 71, "CD3OD": 31, "Not specified": 1, "unknown": 4, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 12, "CDC13": 1, "pyridine-d5": 7, "MeOD": 3, "δ": 1, "chloroform": 1, "dimethylformamide": 1, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "water/methanol": 1, "d": 1, "CHCl3-EtOH (1:1)": 1, "CHCl3:EtOH (1:1)": 1, "acetone-d6": 3, "mixture of two isomers": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 422, "frequency": { "50": 2, "75": 5, "80": 2, "100": 27, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "not specified": 2, "DEPT 135": 1, "CD3OD": 30, "unknown": 2, "CHCl3": 2, "CDCl3": 67, "MeOH": 6, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "δ": 1, "chloroform": 1, "dimethylformamide": 1, "D2O": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "TBD": 1, "acetone-d6": 4 } } }, "yaml": { "parsable": 621, "notParsable": 21, "empty": 2925 }, "numbers": { "inventedNumbers": [ { "number": 400, "counter": 13 }, { "number": 100, "counter": 11 }, { "number": 0, "counter": 10 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 437 }, { "number": -1, "counter": 179 }, { "number": 1, "counter": 78 }, { "number": 14, "counter": 62 }, { "number": 3, "counter": 50 }, { "number": 10, "counter": 48 }, { "number": 4, "counter": 46 }, { "number": 18, "counter": 45 }, { "number": 15, "counter": 44 }, { "number": 16, "counter": 44 }, { "number": 6, "counter": 43 }, { "number": 5, "counter": 42 }, { "number": 8, "counter": 41 }, { "number": 12, "counter": 39 }, { "number": 2, "counter": 38 }, { "number": 9, "counter": 33 }, { "number": 17, "counter": 31 }, { "number": 24, "counter": 27 }, { "number": 11, "counter": 24 }, { "number": 25, "counter": 24 }, { "number": 7, "counter": 23 }, { "number": 23, "counter": 23 }, { "number": 20, "counter": 22 }, { "number": -8, "counter": 22 }, { "number": 19, "counter": 21 }, { "number": 21, "counter": 21 }, { "number": -10, "counter": 21 }, { "number": -7, "counter": 19 }, { "number": 100, "counter": 17 }, { "number": -3, "counter": 15 }, { "number": -2, "counter": 15 }, { "number": -13, "counter": 15 }, { "number": 22, "counter": 14 }, { "number": -4, "counter": 14 }, { "number": 28, "counter": 13 }, { "number": 27, "counter": 11 }, { "number": 30, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": -6, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": -11, "counter": 8 }, { "number": -12, "counter": 8 }, { "number": 29, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 209, "counter": 6 }, { "number": 224, "counter": 6 }, { "number": 230, "counter": 6 }, { "number": 36, "counter": 5 } ], "valid": 499, "invalid": 143, "ratioPresent": 0.8967719935349383, "nbUnion": 35342, "duplicates": 513, "invented": 66 }, "meta": { "withName": 601, "withoutName": 41, "withMolfile": 452 }, "info": { "protonNMR": { "present": 576, "notPresent": 66, "valid": 217, "notValid": 359 }, "carbonNMR": { "present": 423, "notPresent": 219, "valid": 170, "notValid": 253 }, "elementalAnalysis": { "present": 252, "notPresent": 390, "valid": 212, "notValid": 40 }, "hrms": { "present": 86, "notPresent": 556, "valid": 10, "notValid": 76 } }, "logs": { "info": 67, "warn": 201, "error": 184, "fatal": 190 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 468854, "totalTokens": 945591, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "cost": 1.414445 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_0.5_1", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 621 ], "notParsable": [ 21 ], "empty": [ 2925 ] }, "numbers": { "valid": [ 499 ], "invalid": [ 143 ], "ratioPresent": [ 0.8967719935349383 ], "nbUnion": [ 35342 ], "duplicates": [ 513 ], "invented": [ 66 ] }, "meta": { "withName": [ 601 ], "withoutName": [ 41 ], "withMolfile": [ 452 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_0.5_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-1_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-1_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 552, "frequency": { "50": 4, "80": 2, "200": 3, "300": 12, "400": 31, "500": 3, "not specified": 1 }, "solvent": { "CDCl3": 85, "not specified": 36, "DMSO-d6": 67, "not provided": 2, "CD3OD": 29, "unknown": 9, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 5, "MeOH": 6, "CDC13": 1, "Not specified": 10, "pyridine-d5": 6, "MeOD": 3, "N/A": 2, "D2O": 2, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 432, "frequency": { "50": 2, "75": 6, "80": 2, "100": 27, "125": 3, "200": 1, "400": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "CDCl3": 71, "not specified": 25, "CD3OD": 28, "unknown": 8, "CHCl3": 2, "MeOH": 6, "CD3COCD3": 1, "Not specified": 10, "pyridine-d5": 6, "not provided": 1, "CD3CN": 3, "MeOD": 3, "DMSO": 2, "unspecified": 1, "N/A": 2, "D2O": 1, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "CH3OH": 1 } } }, "yaml": { "parsable": 593, "notParsable": 49, "empty": 2400 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 32 }, { "number": 400, "counter": 18 }, { "number": 0, "counter": 7 }, { "number": 15, "counter": 3 }, { "number": 50, "counter": 3 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 11, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 16, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 60, "counter": 1 }, { "number": 75, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": -1, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 193 }, { "number": 1, "counter": 149 }, { "number": 3, "counter": 83 }, { "number": 2, "counter": 72 }, { "number": 4, "counter": 67 }, { "number": 6, "counter": 67 }, { "number": 8, "counter": 61 }, { "number": 5, "counter": 56 }, { "number": 7, "counter": 45 }, { "number": 10, "counter": 43 }, { "number": 14, "counter": 41 }, { "number": 9, "counter": 38 }, { "number": 100, "counter": 37 }, { "number": 12, "counter": 33 }, { "number": 15, "counter": 31 }, { "number": -3, "counter": 31 }, { "number": -2, "counter": 30 }, { "number": 11, "counter": 27 }, { "number": 18, "counter": 27 }, { "number": 17, "counter": 26 }, { "number": 24, "counter": 24 }, { "number": 25, "counter": 22 }, { "number": -4, "counter": 21 }, { "number": 16, "counter": 20 }, { "number": -5, "counter": 19 }, { "number": 23, "counter": 17 }, { "number": 19, "counter": 15 }, { "number": 20, "counter": 15 }, { "number": 22, "counter": 15 }, { "number": -6, "counter": 15 }, { "number": 21, "counter": 14 }, { "number": -8, "counter": 14 }, { "number": -9, "counter": 13 }, { "number": 28, "counter": 11 }, { "number": 30, "counter": 11 }, { "number": 63, "counter": 11 }, { "number": -7, "counter": 11 }, { "number": -10, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 90, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": -11, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 29, "counter": 8 }, { "number": 50, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 224, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": 239, "counter": 7 } ], "valid": 510, "invalid": 132, "ratioPresent": 0.8571060334966574, "nbUnion": 35379, "duplicates": 287, "invented": 105 }, "meta": { "withName": 582, "withoutName": 60, "withMolfile": 453 }, "info": { "protonNMR": { "present": 554, "notPresent": 88, "valid": 194, "notValid": 360 }, "carbonNMR": { "present": 433, "notPresent": 209, "valid": 150, "notValid": 283 }, "elementalAnalysis": { "present": 236, "notPresent": 406, "valid": 202, "notValid": 34 }, "hrms": { "present": 87, "notPresent": 555, "valid": 8, "notValid": 79 } }, "logs": { "info": 60, "warn": 169, "error": 224, "fatal": 189 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 463593, "totalTokens": 937762, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "cost": 1.4013550000000001 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_0.5_1", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 593 ], "notParsable": [ 49 ], "empty": [ 2400 ] }, "numbers": { "valid": [ 510 ], "invalid": [ 132 ], "ratioPresent": [ 0.8571060334966574 ], "nbUnion": [ 35379 ], "duplicates": [ 287 ], "invented": [ 105 ] }, "meta": { "withName": [ 582 ], "withoutName": [ 60 ], "withMolfile": [ 453 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_0.5_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-0.5-1_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-0.5-1_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 589, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 22, "500": 3 }, "solvent": { "CDCl3": 87, "Not specified": 1, "DMSO-d6": 74, "unknown": 3, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CD3OD": 31, "null": 1, "CDC13": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 8, "MeOD": 3, "not specified": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 438, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 2, "200": 2, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 49, "CHCl3": 1, "CDCl3": 70, "CD3OD": 28, "null": 1, "CD3COCD3": 1, "N/A": 2, "pyridine-d5": 7, "MeOH": 6, "unknown": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "not specified": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "CH3OH": 1 } } }, "yaml": { "parsable": 638, "notParsable": 4, "empty": 1941 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 12 }, { "number": 0, "counter": 9 }, { "number": 400, "counter": 4 }, { "number": 50, "counter": 2 }, { "number": 15, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": -18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 7.5, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 433 }, { "number": -1, "counter": 183 }, { "number": 1, "counter": 73 }, { "number": 14, "counter": 54 }, { "number": 4, "counter": 47 }, { "number": 16, "counter": 44 }, { "number": 10, "counter": 43 }, { "number": 15, "counter": 43 }, { "number": 3, "counter": 40 }, { "number": 12, "counter": 39 }, { "number": 18, "counter": 39 }, { "number": 8, "counter": 37 }, { "number": 6, "counter": 36 }, { "number": 9, "counter": 36 }, { "number": 5, "counter": 34 }, { "number": 2, "counter": 30 }, { "number": 17, "counter": 27 }, { "number": 7, "counter": 26 }, { "number": 24, "counter": 24 }, { "number": 25, "counter": 24 }, { "number": 11, "counter": 22 }, { "number": 23, "counter": 22 }, { "number": -10, "counter": 21 }, { "number": -7, "counter": 20 }, { "number": 19, "counter": 19 }, { "number": 21, "counter": 18 }, { "number": 20, "counter": 17 }, { "number": 28, "counter": 16 }, { "number": 100, "counter": 16 }, { "number": -9, "counter": 16 }, { "number": -8, "counter": 16 }, { "number": 22, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 30, "counter": 10 }, { "number": 32, "counter": 10 }, { "number": -2, "counter": 10 }, { "number": -4, "counter": 9 }, { "number": -5, "counter": 8 }, { "number": -13, "counter": 8 }, { "number": 27, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": -3, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 36, "counter": 6 }, { "number": 43, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 267, "counter": 6 }, { "number": 0.1, "counter": 6 }, { "number": 4.2, "counter": 6 } ], "valid": 516, "invalid": 126, "ratioPresent": 0.9073163164922603, "nbUnion": 35334, "duplicates": 295, "invented": 59 }, "meta": { "withName": 619, "withoutName": 23, "withMolfile": 464 }, "info": { "protonNMR": { "present": 589, "notPresent": 53, "valid": 223, "notValid": 366 }, "carbonNMR": { "present": 438, "notPresent": 204, "valid": 174, "notValid": 264 }, "elementalAnalysis": { "present": 248, "notPresent": 394, "valid": 208, "notValid": 40 }, "hrms": { "present": 91, "notPresent": 551, "valid": 10, "notValid": 81 } }, "logs": { "info": 64, "warn": 205, "error": 195, "fatal": 178 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 442273, "totalTokens": 924788, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 0.5, "top_p": 1, "cost": 1.367061 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_0.5_1", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 638 ], "notParsable": [ 4 ], "empty": [ 1941 ] }, "numbers": { "valid": [ 516 ], "invalid": [ 126 ], "ratioPresent": [ 0.9073163164922603 ], "nbUnion": [ 35334 ], "duplicates": [ 295 ], "invented": [ 59 ] }, "meta": { "withName": [ 619 ], "withoutName": [ 23 ], "withMolfile": [ 464 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_1_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-0.5_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-0.5_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 591, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 29, "500": 6 }, "solvent": { "CDCl3": 102, "not specified": 10, "DMSO-d6": 73, "CD3OD": 33, "CD3COCD3": 2, "CD3CN": 6, "null": 1, "unspecified": 2, "CDC13": 1, "N/A": 3, "Not specified": 5, "pyridine-d5": 7, "MeOH": 6, "not provided": 1, "unknown": 3, "MeOD": 3, "Not known": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 436, "frequency": { "50": 2, "75": 5, "80": 2, "100": 27, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 88, "CD3OD": 30, "unspecified": 4, "CD3COCD3": 1, "N/A": 3, "Not specified": 4, "pyridine-d5": 7, "MeOH": 4, "not specified": 6, "not provided": 1, "unknown": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "Not known": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4 } } }, "yaml": { "parsable": 638, "notParsable": 4, "empty": 1897 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 13 }, { "number": 400, "counter": 11 }, { "number": 29, "counter": 3 }, { "number": 0, "counter": 2 }, { "number": 8, "counter": 2 }, { "number": 15, "counter": 2 }, { "number": 36, "counter": 2 }, { "number": 37, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 19, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 28, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 58, "counter": 1 }, { "number": 72, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 3.9, "counter": 1 }, { "number": 5.28, "counter": 1 }, { "number": 11.92, "counter": 1 }, { "number": 64.79, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 3.16, "counter": 1 }, { "number": 79.14, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 433 }, { "number": -1, "counter": 163 }, { "number": 1, "counter": 88 }, { "number": 14, "counter": 56 }, { "number": 16, "counter": 47 }, { "number": 10, "counter": 40 }, { "number": 15, "counter": 40 }, { "number": 8, "counter": 38 }, { "number": 9, "counter": 38 }, { "number": 18, "counter": 38 }, { "number": 5, "counter": 37 }, { "number": 4, "counter": 36 }, { "number": 12, "counter": 34 }, { "number": 6, "counter": 33 }, { "number": 17, "counter": 33 }, { "number": 3, "counter": 32 }, { "number": 2, "counter": 30 }, { "number": 24, "counter": 27 }, { "number": 19, "counter": 24 }, { "number": 25, "counter": 23 }, { "number": 20, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 21, "counter": 21 }, { "number": 23, "counter": 21 }, { "number": 7, "counter": 19 }, { "number": 28, "counter": 17 }, { "number": 100, "counter": 17 }, { "number": 224, "counter": 14 }, { "number": 22, "counter": 13 }, { "number": 211, "counter": 13 }, { "number": 239, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 196, "counter": 10 }, { "number": 267, "counter": 10 }, { "number": -10, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": -7, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 30, "counter": 8 }, { "number": 282, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": -4, "counter": 8 }, { "number": 70, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": -6, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": 251, "counter": 6 } ], "valid": 530, "invalid": 112, "ratioPresent": 0.9143460591758359, "nbUnion": 35379, "duplicates": 190, "invented": 98 }, "meta": { "withName": 636, "withoutName": 6, "withMolfile": 482 }, "info": { "protonNMR": { "present": 592, "notPresent": 50, "valid": 231, "notValid": 361 }, "carbonNMR": { "present": 438, "notPresent": 204, "valid": 172, "notValid": 266 }, "elementalAnalysis": { "present": 265, "notPresent": 377, "valid": 216, "notValid": 49 }, "hrms": { "present": 84, "notPresent": 558, "valid": 12, "notValid": 72 } }, "logs": { "info": 76, "warn": 205, "error": 201, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448941, "totalTokens": 968692, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "cost": 1.417633 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_1_0.5", "version": "v1" } ], "versions": [ "v1" ], "stats": { "yaml": { "parsable": [ 638 ], "notParsable": [ 4 ], "empty": [ 1897 ] }, "numbers": { "valid": [ 530 ], "invalid": [ 112 ], "ratioPresent": [ 0.9143460591758359 ], "nbUnion": [ 35379 ], "duplicates": [ 190 ], "invented": [ 98 ] }, "meta": { "withName": [ 636 ], "withoutName": [ 6 ], "withMolfile": [ 482 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_1_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-0.5_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-0.5_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 568, "frequency": { "50": 4, "80": 2, "200": 3, "300": 10, "400": 49, "500": 7 }, "solvent": { "unknown": 97, "CDCl3": 99, "DMSO-d6": 71, "not specified": 15, "CD3OD": 30, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 5, "MeOH": 7, "CDC13": 1, "N/A": 3, "pyridine-d5": 6, "Not specified": 1, "not available": 1, "methanol": 1, "MeOD": 3, "dimethylformamide": 2, "dioxane": 1, "D2O": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 486, "frequency": { "50": 2, "75": 4, "80": 2, "100": 47, "125": 3, "200": 1, "400": 2, "50.3": 1, "75.4": 1 }, "solvent": { "unknown": 78, "DMSO-d6": 53, "CDCl3": 93, "not specified": 13, "CD3OD": 28, "CHCl3": 1, "MeOH": 4, "CD3COCD3": 1, "N/A": 3, "pyridine-d5": 6, "Not specified": 2, "not available": 1, "methanol": 1, "CD3CN": 3, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 627, "notParsable": 15, "empty": 15601 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 147 }, { "number": 400, "counter": 33 }, { "number": 100, "counter": 11 }, { "number": -18, "counter": 2 }, { "number": 20, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 41, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 435 }, { "number": -1, "counter": 148 }, { "number": 1, "counter": 73 }, { "number": 14, "counter": 51 }, { "number": 4, "counter": 45 }, { "number": 5, "counter": 44 }, { "number": 3, "counter": 41 }, { "number": 100, "counter": 41 }, { "number": 10, "counter": 39 }, { "number": 15, "counter": 38 }, { "number": 8, "counter": 36 }, { "number": 18, "counter": 36 }, { "number": 12, "counter": 35 }, { "number": 17, "counter": 33 }, { "number": 2, "counter": 31 }, { "number": 11, "counter": 31 }, { "number": 16, "counter": 31 }, { "number": 9, "counter": 30 }, { "number": 6, "counter": 27 }, { "number": 24, "counter": 26 }, { "number": 25, "counter": 24 }, { "number": 7, "counter": 22 }, { "number": 19, "counter": 20 }, { "number": 224, "counter": 20 }, { "number": 21, "counter": 18 }, { "number": 28, "counter": 18 }, { "number": 211, "counter": 17 }, { "number": 239, "counter": 17 }, { "number": 23, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": 20, "counter": 15 }, { "number": 196, "counter": 15 }, { "number": 27, "counter": 13 }, { "number": -8, "counter": 13 }, { "number": 22, "counter": 12 }, { "number": 43, "counter": 12 }, { "number": -2, "counter": 12 }, { "number": -3, "counter": 12 }, { "number": 30, "counter": 11 }, { "number": 135, "counter": 11 }, { "number": 254, "counter": 11 }, { "number": 282, "counter": 11 }, { "number": -13, "counter": 11 }, { "number": -10, "counter": 11 }, { "number": -7, "counter": 11 }, { "number": -5, "counter": 11 }, { "number": -4, "counter": 11 }, { "number": 70, "counter": 10 }, { "number": 26, "counter": 8 }, { "number": 29, "counter": 8 } ], "valid": 396, "invalid": 246, "ratioPresent": 0.8948938481387873, "nbUnion": 35496, "duplicates": 271, "invented": 593 }, "meta": { "withName": 619, "withoutName": 23, "withMolfile": 472 }, "info": { "protonNMR": { "present": 586, "notPresent": 56, "valid": 173, "notValid": 413 }, "carbonNMR": { "present": 516, "notPresent": 126, "valid": 152, "notValid": 364 }, "elementalAnalysis": { "present": 275, "notPresent": 367, "valid": 231, "notValid": 44 }, "hrms": { "present": 52, "notPresent": 590, "valid": 9, "notValid": 43 } }, "logs": { "info": 42, "warn": 185, "error": 245, "fatal": 170 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 532172, "totalTokens": 1019181, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "cost": 1.5513530000000002 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_1_0.5", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 627 ], "notParsable": [ 15 ], "empty": [ 15601 ] }, "numbers": { "valid": [ 396 ], "invalid": [ 246 ], "ratioPresent": [ 0.8948938481387873 ], "nbUnion": [ 35496 ], "duplicates": [ 271 ], "invented": [ 593 ] }, "meta": { "withName": [ 619 ], "withoutName": [ 23 ], "withMolfile": [ 472 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_1_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-0.5_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-0.5_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 584, "frequency": { "20": 1, "50": 4, "80": 2, "200": 3, "300": 11, "400": 31, "500": 4 }, "solvent": { "not specified": 5, "N/A": 6, "DMSO-d6": 72, "CDCl3": 85, "unknown": 5, "CD3OD": 31, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 7, "MeOH": 10, "CH2Cl2": 1, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 436, "frequency": { "20": 1, "50": 2, "75": 5, "80": 2, "100": 24, "125": 3, "200": 2, "50.3": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 72, "unknown": 5, "CD3OD": 29, "CHCl3": 2, "CD3COCD3": 1, "N/A": 4, "pyridine-d5": 6, "MeOH": 6, "CH2Cl2": 1, "not specified": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "CH3OH": 1 } } }, "yaml": { "parsable": 627, "notParsable": 15, "empty": 2510 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 16 }, { "number": 0, "counter": 13 }, { "number": 400, "counter": 13 }, { "number": -18, "counter": 2 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": -23.86, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 437 }, { "number": -1, "counter": 177 }, { "number": 1, "counter": 77 }, { "number": 14, "counter": 59 }, { "number": 15, "counter": 47 }, { "number": 3, "counter": 42 }, { "number": 16, "counter": 42 }, { "number": 4, "counter": 41 }, { "number": 10, "counter": 41 }, { "number": 18, "counter": 41 }, { "number": 2, "counter": 37 }, { "number": 5, "counter": 37 }, { "number": 9, "counter": 36 }, { "number": 12, "counter": 36 }, { "number": 6, "counter": 35 }, { "number": 24, "counter": 33 }, { "number": 17, "counter": 32 }, { "number": 8, "counter": 31 }, { "number": -10, "counter": 28 }, { "number": 20, "counter": 27 }, { "number": -7, "counter": 25 }, { "number": 11, "counter": 22 }, { "number": 19, "counter": 22 }, { "number": 21, "counter": 21 }, { "number": 23, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": -8, "counter": 20 }, { "number": -13, "counter": 18 }, { "number": 100, "counter": 17 }, { "number": 22, "counter": 16 }, { "number": -9, "counter": 16 }, { "number": 26, "counter": 13 }, { "number": -3, "counter": 12 }, { "number": -12, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 28, "counter": 10 }, { "number": -2, "counter": 10 }, { "number": -4, "counter": 10 }, { "number": 30, "counter": 9 }, { "number": 33, "counter": 9 }, { "number": -6, "counter": 8 }, { "number": 224, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": 32, "counter": 6 }, { "number": 36, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 0.1, "counter": 6 }, { "number": -14, "counter": 6 } ], "valid": 506, "invalid": 136, "ratioPresent": 0.9051599564242956, "nbUnion": 35347, "duplicates": 341, "invented": 86 }, "meta": { "withName": 612, "withoutName": 30, "withMolfile": 461 }, "info": { "protonNMR": { "present": 585, "notPresent": 57, "valid": 228, "notValid": 357 }, "carbonNMR": { "present": 437, "notPresent": 205, "valid": 172, "notValid": 265 }, "elementalAnalysis": { "present": 253, "notPresent": 389, "valid": 214, "notValid": 39 }, "hrms": { "present": 85, "notPresent": 557, "valid": 11, "notValid": 74 } }, "logs": { "info": 67, "warn": 216, "error": 178, "fatal": 181 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 466312, "totalTokens": 943049, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "cost": 1.409361 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_1_0.5", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 627 ], "notParsable": [ 15 ], "empty": [ 2510 ] }, "numbers": { "valid": [ 506 ], "invalid": [ 136 ], "ratioPresent": [ 0.9051599564242956 ], "nbUnion": [ 35347 ], "duplicates": [ 341 ], "invented": [ 86 ] }, "meta": { "withName": [ 612 ], "withoutName": [ 30 ], "withMolfile": [ 461 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_1_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-0.5_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-0.5_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 561, "frequency": { "50": 4, "80": 2, "200": 3, "300": 10, "400": 37, "500": 3, "Not specified": 1, "12.62": 1 }, "solvent": { "CDCl3": 86, "not specified": 33, "not provided": 2, "CD3OD": 33, "Not specified": 13, "CHCl3": 2, "DMSO-d6": 67, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 7, "unknown": 8, "CDC13": 1, "pyridine-d5": 6, "MeOD": 3, "N/A": 1, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "δ": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 436, "frequency": { "50": 2, "75": 5, "80": 2, "100": 31, "125": 3, "200": 1, "400": 1, "50.3": 1, "75.4": 1, "156.2": 1 }, "solvent": { "DMSO-d6": 49, "not specified": 22, "DEPT 135": 1, "CD3OD": 31, "CDCl3": 74, "CHCl3": 3, "MeOH": 7, "unknown": 6, "CD3COCD3": 1, "Not specified": 9, "pyridine-d5": 5, "not provided": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "N/A": 1, "D2O": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "δ": 1, "CH3OH": 1 } } }, "yaml": { "parsable": 598, "notParsable": 44, "empty": 2053 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 30 }, { "number": 400, "counter": 20 }, { "number": 0, "counter": 10 }, { "number": 50, "counter": 7 }, { "number": 7, "counter": 2 }, { "number": 14, "counter": 2 }, { "number": 15, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 6, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 10, "counter": 1 }, { "number": 11, "counter": 1 }, { "number": 12, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 21, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": -11, "counter": 1 }, { "number": -10, "counter": 1 }, { "number": -9, "counter": 1 }, { "number": -8, "counter": 1 }, { "number": -7, "counter": 1 }, { "number": -6, "counter": 1 }, { "number": -5, "counter": 1 }, { "number": 3.83, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 197 }, { "number": 1, "counter": 141 }, { "number": 3, "counter": 77 }, { "number": 2, "counter": 64 }, { "number": 6, "counter": 63 }, { "number": 4, "counter": 59 }, { "number": 5, "counter": 55 }, { "number": 8, "counter": 49 }, { "number": 12, "counter": 43 }, { "number": 10, "counter": 41 }, { "number": 14, "counter": 40 }, { "number": 7, "counter": 39 }, { "number": 9, "counter": 35 }, { "number": 100, "counter": 35 }, { "number": 18, "counter": 34 }, { "number": -3, "counter": 30 }, { "number": 15, "counter": 29 }, { "number": 17, "counter": 26 }, { "number": 11, "counter": 23 }, { "number": -2, "counter": 23 }, { "number": 24, "counter": 22 }, { "number": -4, "counter": 22 }, { "number": 16, "counter": 20 }, { "number": 25, "counter": 20 }, { "number": 20, "counter": 19 }, { "number": 23, "counter": 18 }, { "number": 19, "counter": 16 }, { "number": 28, "counter": 15 }, { "number": 21, "counter": 14 }, { "number": 50, "counter": 12 }, { "number": 22, "counter": 11 }, { "number": 70, "counter": 11 }, { "number": 63, "counter": 10 }, { "number": 90, "counter": 10 }, { "number": -5, "counter": 10 }, { "number": 91, "counter": 9 }, { "number": 237, "counter": 9 }, { "number": 360, "counter": 9 }, { "number": -6, "counter": 9 }, { "number": 27, "counter": 7 }, { "number": 34, "counter": 7 }, { "number": 77, "counter": 7 }, { "number": 225, "counter": 7 }, { "number": -10, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": -7, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 26, "counter": 6 }, { "number": 30, "counter": 6 } ], "valid": 495, "invalid": 147, "ratioPresent": 0.8682782424720836, "nbUnion": 35399, "duplicates": 248, "invented": 131 }, "meta": { "withName": 584, "withoutName": 58, "withMolfile": 442 }, "info": { "protonNMR": { "present": 562, "notPresent": 80, "valid": 192, "notValid": 370 }, "carbonNMR": { "present": 436, "notPresent": 206, "valid": 157, "notValid": 279 }, "elementalAnalysis": { "present": 232, "notPresent": 410, "valid": 193, "notValid": 39 }, "hrms": { "present": 102, "notPresent": 540, "valid": 9, "notValid": 93 } }, "logs": { "info": 61, "warn": 147, "error": 234, "fatal": 200 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 465133, "totalTokens": 939302, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "cost": 1.404435 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_1_0.5", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 598 ], "notParsable": [ 44 ], "empty": [ 2053 ] }, "numbers": { "valid": [ 495 ], "invalid": [ 147 ], "ratioPresent": [ 0.8682782424720836 ], "nbUnion": [ 35399 ], "duplicates": [ 248 ], "invented": [ 131 ] }, "meta": { "withName": [ 584 ], "withoutName": [ 58 ], "withMolfile": [ 442 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_1_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-0.5_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-0.5_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 588, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 23, "500": 5, "16.9": 1 }, "solvent": { "CDCl3": 85, "not specified": 1, "DMSO-d6": 72, "CD3OD": 32, "unknown": 4, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 10, "CDC13": 1, "pyridine-d5": 7, "MeOD": 3, "qz": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 434, "frequency": { "50": 2, "75": 5, "80": 2, "100": 21, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1, "16.9": 1 }, "solvent": { "DMSO-d6": 49, "CD3OD": 29, "CHCl3": 1, "CDCl3": 67, "CD3COCD3": 1, "unknown": 1, "pyridine-d5": 7, "MeOH": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "ppm": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 1924 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 12 }, { "number": 0, "counter": 10 }, { "number": 400, "counter": 5 }, { "number": -18, "counter": 2 }, { "number": 34, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 435 }, { "number": -1, "counter": 181 }, { "number": 1, "counter": 73 }, { "number": 14, "counter": 50 }, { "number": 4, "counter": 44 }, { "number": 15, "counter": 42 }, { "number": 10, "counter": 41 }, { "number": 3, "counter": 38 }, { "number": 8, "counter": 38 }, { "number": 16, "counter": 38 }, { "number": 18, "counter": 38 }, { "number": 6, "counter": 37 }, { "number": 5, "counter": 35 }, { "number": 9, "counter": 35 }, { "number": 12, "counter": 33 }, { "number": 17, "counter": 32 }, { "number": 2, "counter": 25 }, { "number": 24, "counter": 25 }, { "number": 7, "counter": 23 }, { "number": 11, "counter": 23 }, { "number": 23, "counter": 23 }, { "number": 25, "counter": 22 }, { "number": -10, "counter": 21 }, { "number": 19, "counter": 20 }, { "number": 20, "counter": 16 }, { "number": 21, "counter": 16 }, { "number": 28, "counter": 16 }, { "number": -9, "counter": 16 }, { "number": -7, "counter": 16 }, { "number": 100, "counter": 12 }, { "number": -2, "counter": 12 }, { "number": 22, "counter": 11 }, { "number": -8, "counter": 11 }, { "number": 30, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": -4, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 239, "counter": 9 }, { "number": -13, "counter": 9 }, { "number": 70, "counter": 8 }, { "number": 196, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": -3, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 240, "counter": 7 }, { "number": 254, "counter": 7 }, { "number": -12, "counter": 7 }, { "number": 0, "counter": 6 } ], "valid": 537, "invalid": 105, "ratioPresent": 0.9155303112600338, "nbUnion": 35331, "duplicates": 247, "invented": 61 }, "meta": { "withName": 616, "withoutName": 26, "withMolfile": 459 }, "info": { "protonNMR": { "present": 588, "notPresent": 54, "valid": 228, "notValid": 360 }, "carbonNMR": { "present": 434, "notPresent": 208, "valid": 170, "notValid": 264 }, "elementalAnalysis": { "present": 244, "notPresent": 398, "valid": 206, "notValid": 38 }, "hrms": { "present": 79, "notPresent": 563, "valid": 10, "notValid": 69 } }, "logs": { "info": 71, "warn": 212, "error": 176, "fatal": 183 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 443042, "totalTokens": 925557, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 0.5, "cost": 1.368599 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_1_0.5", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 639 ], "notParsable": [ 3 ], "empty": [ 1924 ] }, "numbers": { "valid": [ 537 ], "invalid": [ 105 ], "ratioPresent": [ 0.9155303112600338 ], "nbUnion": [ 35331 ], "duplicates": [ 247 ], "invented": [ 61 ] }, "meta": { "withName": [ 616 ], "withoutName": [ 26 ], "withMolfile": [ 459 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_1_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-1_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-1_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 565, "frequency": { "50": 3, "80": 2, "200": 2, "300": 11, "400": 21, "500": 4, "N/A": 1 }, "solvent": { "CDCl3": 81, "not specified": 3, "DMSO-d6": 70, "unknown": 3, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "CD3OD": 30, "N/A": 1, "pyridine-d5": 7, "MeOH": 5, "MeOD": 3, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "-": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 405, "frequency": { "50": 2, "75": 5, "80": 2, "100": 18, "125": 2, "N/A": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 47, "unknown": 2, "CDCl3": 63, "not specified": 1, "CHCl3": 2, "CD3OD": 29, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 6, "MeOH": 3, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 622, "notParsable": 20, "empty": 2071 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 16 }, { "number": 0, "counter": 9 }, { "number": 400, "counter": 4 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 7.5, "counter": 2 }, { "number": 15, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 75, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": -23.86, "counter": 1 }, { "number": 8.6, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 115.4, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 428 }, { "number": -1, "counter": 205 }, { "number": 1, "counter": 109 }, { "number": 4, "counter": 68 }, { "number": 3, "counter": 55 }, { "number": 10, "counter": 52 }, { "number": 5, "counter": 51 }, { "number": 8, "counter": 51 }, { "number": 14, "counter": 51 }, { "number": 6, "counter": 50 }, { "number": 2, "counter": 49 }, { "number": 16, "counter": 49 }, { "number": 15, "counter": 42 }, { "number": 12, "counter": 39 }, { "number": -8, "counter": 36 }, { "number": 9, "counter": 35 }, { "number": 18, "counter": 35 }, { "number": -7, "counter": 33 }, { "number": 7, "counter": 32 }, { "number": 17, "counter": 32 }, { "number": -10, "counter": 32 }, { "number": -4, "counter": 29 }, { "number": -3, "counter": 28 }, { "number": -2, "counter": 27 }, { "number": 24, "counter": 26 }, { "number": -9, "counter": 26 }, { "number": 25, "counter": 25 }, { "number": 11, "counter": 23 }, { "number": -13, "counter": 23 }, { "number": 19, "counter": 22 }, { "number": 20, "counter": 20 }, { "number": 21, "counter": 20 }, { "number": 100, "counter": 20 }, { "number": -5, "counter": 19 }, { "number": 23, "counter": 18 }, { "number": 28, "counter": 17 }, { "number": -6, "counter": 17 }, { "number": -12, "counter": 16 }, { "number": 22, "counter": 11 }, { "number": -14, "counter": 11 }, { "number": 30, "counter": 10 }, { "number": 211, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": 70, "counter": 9 }, { "number": -11, "counter": 9 }, { "number": 26, "counter": 8 }, { "number": 135, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 29, "counter": 7 } ], "valid": 536, "invalid": 106, "ratioPresent": 0.8840547148832081, "nbUnion": 35339, "duplicates": 162, "invented": 68 }, "meta": { "withName": 596, "withoutName": 46, "withMolfile": 450 }, "info": { "protonNMR": { "present": 567, "notPresent": 75, "valid": 210, "notValid": 357 }, "carbonNMR": { "present": 408, "notPresent": 234, "valid": 163, "notValid": 245 }, "elementalAnalysis": { "present": 243, "notPresent": 399, "valid": 204, "notValid": 39 }, "hrms": { "present": 59, "notPresent": 583, "valid": 8, "notValid": 51 } }, "logs": { "info": 61, "warn": 208, "error": 181, "fatal": 192 }, "chatgpt": { "promptTokens": 488935, "completionTokens": 439200, "totalTokens": 928135, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "cost": 1.367335 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_1_1", "version": "v1" } ], "versions": [ "v1" ], "stats": { "yaml": { "parsable": [ 622 ], "notParsable": [ 20 ], "empty": [ 2071 ] }, "numbers": { "valid": [ 536 ], "invalid": [ 106 ], "ratioPresent": [ 0.8840547148832081 ], "nbUnion": [ 35339 ], "duplicates": [ 162 ], "invented": [ 68 ] }, "meta": { "withName": [ 596 ], "withoutName": [ 46 ], "withMolfile": [ 450 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_1_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-1_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-1_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 525, "frequency": { "50": 4, "80": 2, "200": 2, "300": 12, "400": 46, "500": 7, "N/A": 4, "NA": 1, "not specified": 1 }, "solvent": { "CDCl3": 87, "not provided": 3, "unknown": 53, "not specified": 11, "N/A": 12, "Not specified": 7, "CHCl3": 1, "DMSO-d6": 66, "CD3COCD3": 2, "CD3CN": 6, "CD3OD": 26, "Not Specified": 1, "literature": 2, "pyridine-d5": 7, "δ": 1, "MeOH": 7, "Not mentioned": 2, "MeOD": 3, "NA": 1, " ": 1, "not_specified": 1, "dimethylformamide": 1, "dioxane": 1, "D2O": 4, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 2, "Unknown": 1 } }, "carbon": { "nbSpectra": 437, "frequency": { "50": 2, "75": 7, "80": 2, "100": 40, "125": 3, "400": 4, "N/A": 2, "75.4": 1, "NA": 1, "not specified": 1 }, "solvent": { "DMSO-d6": 47, "CDCl3": 75, "not specified": 8, "unknown": 43, "Not specified": 7, "not provided": 2, "CHCl3": 1, "CD3OD": 23, "CD3COCD3": 1, "N/A": 8, "Not Specified": 1, "literature": 2, "pyridine-d5": 7, "δ": 1, "MeOH": 5, "Not mentioned": 2, "CD3CN": 4, "MeOD": 3, "NA": 1, "DMSO": 2, "not_specified": 1, "unspecified": 1, "d4-MeOH": 32, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "Unknown": 1 } } }, "yaml": { "parsable": 595, "notParsable": 47, "empty": 14218 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 150 }, { "number": 400, "counter": 33 }, { "number": 100, "counter": 18 }, { "number": 3, "counter": 2 }, { "number": 75, "counter": 2 }, { "number": 300, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 9, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 55, "counter": 1 }, { "number": 101, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 125, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 500, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1207, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.02, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 7.18, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 446 }, { "number": -1, "counter": 187 }, { "number": 1, "counter": 113 }, { "number": 4, "counter": 75 }, { "number": 14, "counter": 70 }, { "number": 3, "counter": 69 }, { "number": 2, "counter": 64 }, { "number": 5, "counter": 60 }, { "number": 8, "counter": 54 }, { "number": 15, "counter": 52 }, { "number": 6, "counter": 51 }, { "number": 18, "counter": 45 }, { "number": 12, "counter": 44 }, { "number": 9, "counter": 42 }, { "number": 10, "counter": 41 }, { "number": 100, "counter": 41 }, { "number": -2, "counter": 40 }, { "number": 17, "counter": 39 }, { "number": 16, "counter": 38 }, { "number": -8, "counter": 34 }, { "number": -4, "counter": 34 }, { "number": -6, "counter": 32 }, { "number": 11, "counter": 31 }, { "number": -3, "counter": 31 }, { "number": -5, "counter": 30 }, { "number": -7, "counter": 29 }, { "number": 7, "counter": 28 }, { "number": 19, "counter": 28 }, { "number": 20, "counter": 27 }, { "number": 25, "counter": 27 }, { "number": 23, "counter": 26 }, { "number": 24, "counter": 26 }, { "number": -10, "counter": 25 }, { "number": -13, "counter": 23 }, { "number": 21, "counter": 21 }, { "number": 22, "counter": 20 }, { "number": -9, "counter": 20 }, { "number": -12, "counter": 18 }, { "number": -14, "counter": 17 }, { "number": 28, "counter": 15 }, { "number": 224, "counter": 15 }, { "number": 30, "counter": 14 }, { "number": 211, "counter": 14 }, { "number": 26, "counter": 13 }, { "number": 43, "counter": 13 }, { "number": 196, "counter": 13 }, { "number": 239, "counter": 13 }, { "number": 135, "counter": 12 }, { "number": 7.9, "counter": 12 }, { "number": -11, "counter": 12 } ], "valid": 373, "invalid": 269, "ratioPresent": 0.8347837357547602, "nbUnion": 35520, "duplicates": 432, "invented": 652 }, "meta": { "withName": 578, "withoutName": 64, "withMolfile": 443 }, "info": { "protonNMR": { "present": 544, "notPresent": 98, "valid": 154, "notValid": 390 }, "carbonNMR": { "present": 479, "notPresent": 163, "valid": 144, "notValid": 335 }, "elementalAnalysis": { "present": 255, "notPresent": 387, "valid": 208, "notValid": 47 }, "hrms": { "present": 54, "notPresent": 588, "valid": 7, "notValid": 47 } }, "logs": { "info": 35, "warn": 153, "error": 254, "fatal": 200 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 528764, "totalTokens": 1015773, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "cost": 1.544537 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_1_1", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 595 ], "notParsable": [ 47 ], "empty": [ 14218 ] }, "numbers": { "valid": [ 373 ], "invalid": [ 269 ], "ratioPresent": [ 0.8347837357547602 ], "nbUnion": [ 35520 ], "duplicates": [ 432 ], "invented": [ 652 ] }, "meta": { "withName": [ 578 ], "withoutName": [ 64 ], "withMolfile": [ 443 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_1_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-1_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-1_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 545, "frequency": { "50": 3, "80": 2, "200": 3, "300": 11, "400": 27, "500": 4, "N/A": 3 }, "solvent": { "CDCl3": 83, "Not specified": 3, "DMSO-d6": 67, "not specified": 2, "unknown": 7, "CD3OD": 26, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 5, "MeOH": 9, "CDC13": 1, "N/A": 4, "pyridine-d5": 7, "literature": 1, "MeOD": 2, "δ": 1, "n-hexane/ethyl acetate": 1, "D2O": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "d": 2, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 405, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "N/A": 2, "50.3": 1, "100.6": 1 }, "solvent": { "DMSO-d6": 46, "not specified": 1, "unknown": 2, "DEPT 135": 1, "CD3OD": 25, "CHCl3": 2, "CDCl3": 69, "CD3COCD3": 1, "N/A": 3, "pyridine-d5": 7, "MeOH": 6, "CD3CN": 3, "MeOD": 2, "DMSO": 1, "δ": 1, "n-hexane/ethyl acetate": 1, "d4-MeOH": 33, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "CH3OH": 1 } } }, "yaml": { "parsable": 590, "notParsable": 52, "empty": 3619 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 26 }, { "number": 100, "counter": 16 }, { "number": 400, "counter": 11 }, { "number": 18, "counter": 2 }, { "number": 27, "counter": 2 }, { "number": 30, "counter": 2 }, { "number": 41, "counter": 2 }, { "number": 58, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 2, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 17, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 48, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 75, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 116, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 456, "counter": 1 }, { "number": 531, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 129.9, "counter": 1 }, { "number": -23.86, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 7.5, "counter": 1 }, { "number": -1, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 443 }, { "number": -1, "counter": 194 }, { "number": 1, "counter": 119 }, { "number": 4, "counter": 79 }, { "number": 2, "counter": 71 }, { "number": 3, "counter": 69 }, { "number": 5, "counter": 66 }, { "number": 6, "counter": 58 }, { "number": 14, "counter": 57 }, { "number": 10, "counter": 56 }, { "number": 8, "counter": 49 }, { "number": 15, "counter": 49 }, { "number": 16, "counter": 45 }, { "number": 9, "counter": 44 }, { "number": 18, "counter": 44 }, { "number": 7, "counter": 40 }, { "number": -2, "counter": 39 }, { "number": 17, "counter": 38 }, { "number": -3, "counter": 38 }, { "number": 12, "counter": 36 }, { "number": 100, "counter": 36 }, { "number": -4, "counter": 35 }, { "number": 11, "counter": 32 }, { "number": 20, "counter": 30 }, { "number": 19, "counter": 28 }, { "number": 24, "counter": 28 }, { "number": -8, "counter": 28 }, { "number": 21, "counter": 26 }, { "number": -5, "counter": 26 }, { "number": 25, "counter": 25 }, { "number": -9, "counter": 25 }, { "number": -10, "counter": 25 }, { "number": 23, "counter": 23 }, { "number": -6, "counter": 23 }, { "number": -7, "counter": 23 }, { "number": 22, "counter": 18 }, { "number": 28, "counter": 18 }, { "number": -13, "counter": 17 }, { "number": -12, "counter": 14 }, { "number": -11, "counter": 14 }, { "number": 26, "counter": 13 }, { "number": 27, "counter": 13 }, { "number": -14, "counter": 13 }, { "number": 7.3, "counter": 11 }, { "number": 36, "counter": 10 }, { "number": 70, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": 30, "counter": 9 }, { "number": 33, "counter": 9 }, { "number": 29, "counter": 8 } ], "valid": 476, "invalid": 166, "ratioPresent": 0.8401865277680249, "nbUnion": 35386, "duplicates": 450, "invented": 138 }, "meta": { "withName": 557, "withoutName": 85, "withMolfile": 416 }, "info": { "protonNMR": { "present": 546, "notPresent": 96, "valid": 187, "notValid": 359 }, "carbonNMR": { "present": 405, "notPresent": 237, "valid": 160, "notValid": 245 }, "elementalAnalysis": { "present": 219, "notPresent": 423, "valid": 187, "notValid": 32 }, "hrms": { "present": 64, "notPresent": 578, "valid": 7, "notValid": 57 } }, "logs": { "info": 55, "warn": 183, "error": 178, "fatal": 226 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 460719, "totalTokens": 937456, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "cost": 1.398175 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_1_1", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 590 ], "notParsable": [ 52 ], "empty": [ 3619 ] }, "numbers": { "valid": [ 476 ], "invalid": [ 166 ], "ratioPresent": [ 0.8401865277680249 ], "nbUnion": [ 35386 ], "duplicates": [ 450 ], "invented": [ 138 ] }, "meta": { "withName": [ 557 ], "withoutName": [ 85 ], "withMolfile": [ 416 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_1_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-1_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-1_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 546, "frequency": { "50": 4, "80": 2, "200": 3, "300": 9, "400": 37, "500": 3, "not specified": 2, "uncertain": 1 }, "solvent": { "CDCl3": 83, "not specified": 26, "unknown": 15, "Not specified": 12, "N/A": 2, "not provided": 4, "CD3OD": 29, "CHCl3": 1, "DMSO-d6": 68, "CD3COCD3": 2, "MeOH": 7, "CDC13": 1, "Not provided": 3, "not_available": 1, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "not stated": 1, "D2O": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "Not available": 1, "uncertain": 1, "acetone-d6": 2, "not_given": 1, "not mentioned": 1 } }, "carbon": { "nbSpectra": 415, "frequency": { "50": 2, "75": 6, "80": 2, "100": 31, "125": 3, "200": 1, "400": 1, "not specified": 1, "50.3": 1, "75.4": 1, "75.47": 1 }, "solvent": { "DMSO-d6": 49, "unknown": 8, "CDCl3": 65, "N/A": 2, "not provided": 4, "not specified": 22, "CD3OD": 28, "CHCl3": 1, "MeOH": 7, "CD3COCD3": 1, "Not provided": 3, "Not specified": 5, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "not stated": 1, "d4-MeOH": 33, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "not_given": 1, "not mentioned": 1 } } }, "yaml": { "parsable": 588, "notParsable": 54, "empty": 2306 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 28 }, { "number": 400, "counter": 21 }, { "number": 0, "counter": 12 }, { "number": 15, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": -3, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 7, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 19, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 43, "counter": 1 }, { "number": 48, "counter": 1 }, { "number": 52, "counter": 1 }, { "number": 76, "counter": 1 }, { "number": 89, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 300, "counter": 1 }, { "number": 475, "counter": 1 }, { "number": 485, "counter": 1 }, { "number": 1000, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 2365, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": -1, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 75.47, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 435 }, { "number": -1, "counter": 206 }, { "number": 1, "counter": 153 }, { "number": 3, "counter": 82 }, { "number": 2, "counter": 75 }, { "number": 4, "counter": 75 }, { "number": 6, "counter": 67 }, { "number": 5, "counter": 64 }, { "number": 8, "counter": 61 }, { "number": 10, "counter": 49 }, { "number": 12, "counter": 39 }, { "number": 14, "counter": 38 }, { "number": 7, "counter": 37 }, { "number": 9, "counter": 37 }, { "number": -3, "counter": 33 }, { "number": 100, "counter": 31 }, { "number": -4, "counter": 31 }, { "number": 18, "counter": 29 }, { "number": 11, "counter": 28 }, { "number": 15, "counter": 28 }, { "number": -2, "counter": 28 }, { "number": 17, "counter": 27 }, { "number": 16, "counter": 24 }, { "number": -5, "counter": 24 }, { "number": 24, "counter": 22 }, { "number": 25, "counter": 22 }, { "number": 20, "counter": 20 }, { "number": -7, "counter": 17 }, { "number": -8, "counter": 16 }, { "number": -6, "counter": 16 }, { "number": 23, "counter": 15 }, { "number": 21, "counter": 13 }, { "number": 28, "counter": 13 }, { "number": -13, "counter": 13 }, { "number": -10, "counter": 13 }, { "number": 19, "counter": 12 }, { "number": 22, "counter": 12 }, { "number": -12, "counter": 11 }, { "number": -11, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 70, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": -14, "counter": 10 }, { "number": -9, "counter": 10 }, { "number": 30, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": 50, "counter": 8 }, { "number": 63, "counter": 8 }, { "number": 1.4, "counter": 8 }, { "number": 7.3, "counter": 8 } ], "valid": 481, "invalid": 161, "ratioPresent": 0.8475616002437726, "nbUnion": 35394, "duplicates": 305, "invented": 127 }, "meta": { "withName": 569, "withoutName": 73, "withMolfile": 430 }, "info": { "protonNMR": { "present": 546, "notPresent": 96, "valid": 183, "notValid": 363 }, "carbonNMR": { "present": 415, "notPresent": 227, "valid": 139, "notValid": 276 }, "elementalAnalysis": { "present": 226, "notPresent": 416, "valid": 186, "notValid": 40 }, "hrms": { "present": 89, "notPresent": 553, "valid": 3, "notValid": 86 } }, "logs": { "info": 55, "warn": 157, "error": 218, "fatal": 212 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 461775, "totalTokens": 935944, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "cost": 1.397719 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_1_1", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 588 ], "notParsable": [ 54 ], "empty": [ 2306 ] }, "numbers": { "valid": [ 481 ], "invalid": [ 161 ], "ratioPresent": [ 0.8475616002437726 ], "nbUnion": [ 35394 ], "duplicates": [ 305 ], "invented": [ 127 ] }, "meta": { "withName": [ 569 ], "withoutName": [ 73 ], "withMolfile": [ 430 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_1_1", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1-1_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1-1_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 580, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 24, "500": 3 }, "solvent": { "CDCl3": 87, "unknown": 4, "DMSO-d6": 71, "CD3OD": 33, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 7, "MeOH": 8, "not provided": 1, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "D2O": 1, "acetone-d6": 3, "δ": 1, "Not provided": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 411, "frequency": { "50": 2, "75": 5, "80": 2, "100": 19, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 48, "CD3OD": 30, "CDCl3": 70, "CHCl3": 1, "CD3COCD3": 1, "pyridine-d5": 7, "MeOH": 5, "not provided": 1, "unknown": 2, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "unspecified": 1, "δ": 1 } } }, "yaml": { "parsable": 629, "notParsable": 13, "empty": 2149 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 13 }, { "number": 0, "counter": 12 }, { "number": 400, "counter": 7 }, { "number": 50, "counter": 3 }, { "number": 31, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 70, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 377, "counter": 1 }, { "number": 529, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": -18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 7.2, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 7.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 441 }, { "number": -1, "counter": 188 }, { "number": 1, "counter": 89 }, { "number": 14, "counter": 53 }, { "number": 3, "counter": 52 }, { "number": 4, "counter": 50 }, { "number": 5, "counter": 47 }, { "number": 6, "counter": 45 }, { "number": 10, "counter": 44 }, { "number": 15, "counter": 42 }, { "number": 18, "counter": 40 }, { "number": 16, "counter": 39 }, { "number": 2, "counter": 36 }, { "number": 12, "counter": 36 }, { "number": 8, "counter": 35 }, { "number": 9, "counter": 33 }, { "number": 17, "counter": 31 }, { "number": -10, "counter": 30 }, { "number": 7, "counter": 27 }, { "number": 24, "counter": 26 }, { "number": 100, "counter": 26 }, { "number": -8, "counter": 25 }, { "number": -7, "counter": 25 }, { "number": 11, "counter": 24 }, { "number": 19, "counter": 24 }, { "number": 20, "counter": 23 }, { "number": 25, "counter": 23 }, { "number": 21, "counter": 22 }, { "number": 23, "counter": 22 }, { "number": -9, "counter": 22 }, { "number": -3, "counter": 22 }, { "number": -4, "counter": 20 }, { "number": -2, "counter": 20 }, { "number": 28, "counter": 18 }, { "number": -13, "counter": 17 }, { "number": -5, "counter": 16 }, { "number": 22, "counter": 15 }, { "number": 30, "counter": 13 }, { "number": -6, "counter": 12 }, { "number": 26, "counter": 11 }, { "number": -11, "counter": 11 }, { "number": -12, "counter": 11 }, { "number": 27, "counter": 10 }, { "number": 32, "counter": 10 }, { "number": 224, "counter": 10 }, { "number": -14, "counter": 10 }, { "number": 211, "counter": 9 }, { "number": 239, "counter": 8 }, { "number": -19, "counter": 8 }, { "number": 31, "counter": 7 } ], "valid": 526, "invalid": 116, "ratioPresent": 0.8878753288742018, "nbUnion": 35343, "duplicates": 254, "invented": 77 }, "meta": { "withName": 609, "withoutName": 33, "withMolfile": 454 }, "info": { "protonNMR": { "present": 581, "notPresent": 61, "valid": 199, "notValid": 382 }, "carbonNMR": { "present": 411, "notPresent": 231, "valid": 159, "notValid": 252 }, "elementalAnalysis": { "present": 242, "notPresent": 400, "valid": 205, "notValid": 37 }, "hrms": { "present": 76, "notPresent": 566, "valid": 10, "notValid": 66 } }, "logs": { "info": 65, "warn": 202, "error": 187, "fatal": 188 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 438249, "totalTokens": 920764, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 1, "top_p": 1, "cost": 1.359013 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_1_1", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 629 ], "notParsable": [ 13 ], "empty": [ 2149 ] }, "numbers": { "valid": [ 526 ], "invalid": [ 116 ], "ratioPresent": [ 0.8878753288742018 ], "nbUnion": [ 35343 ], "duplicates": [ 254 ], "invented": [ 77 ] }, "meta": { "withName": [ 609 ], "withoutName": [ 33 ], "withMolfile": [ 454 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_1.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1.5-0.5_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1.5-0.5_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 595, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 25, "500": 5 }, "solvent": { "CDCl3": 98, "not specified": 8, "unspecified": 2, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "CD3OD": 30, "null": 3, "unknown": 6, "CDC13": 1, "Not specified": 3, "N/A": 2, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 430, "frequency": { "50": 2, "75": 5, "80": 2, "100": 24, "125": 2, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "unspecified": 2, "CDCl3": 79, "CD3OD": 27, "null": 2, "unknown": 4, "CD3COCD3": 1, "Not specified": 3, "N/A": 2, "pyridine-d5": 7, "MeOH": 6, "not specified": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 638, "notParsable": 4, "empty": 1930 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 13 }, { "number": 400, "counter": 7 }, { "number": 0, "counter": 5 }, { "number": 15, "counter": 4 }, { "number": 20, "counter": 2 }, { "number": 21, "counter": 2 }, { "number": 29, "counter": 2 }, { "number": 31, "counter": 2 }, { "number": 32, "counter": 2 }, { "number": 43, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 8.95, "counter": 2 }, { "number": 77.09, "counter": 2 }, { "number": 3, "counter": 1 }, { "number": 4, "counter": 1 }, { "number": 7, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 19, "counter": 1 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 36, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 50, "counter": 1 }, { "number": 56, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 4.15, "counter": 1 }, { "number": 5.29, "counter": 1 }, { "number": 5.36, "counter": 1 }, { "number": 20.79, "counter": 1 }, { "number": 20.94, "counter": 1 }, { "number": 68.62, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 159 }, { "number": 1, "counter": 86 }, { "number": 14, "counter": 53 }, { "number": 10, "counter": 45 }, { "number": 15, "counter": 39 }, { "number": 8, "counter": 38 }, { "number": 5, "counter": 37 }, { "number": 6, "counter": 37 }, { "number": 9, "counter": 37 }, { "number": 18, "counter": 37 }, { "number": 16, "counter": 36 }, { "number": 4, "counter": 33 }, { "number": 12, "counter": 33 }, { "number": 3, "counter": 32 }, { "number": 2, "counter": 28 }, { "number": 11, "counter": 26 }, { "number": 17, "counter": 26 }, { "number": 7, "counter": 25 }, { "number": 21, "counter": 19 }, { "number": 23, "counter": 19 }, { "number": 24, "counter": 19 }, { "number": 25, "counter": 19 }, { "number": 19, "counter": 17 }, { "number": 28, "counter": 17 }, { "number": 20, "counter": 15 }, { "number": 22, "counter": 13 }, { "number": 26, "counter": 10 }, { "number": -7, "counter": 10 }, { "number": -6, "counter": 10 }, { "number": 32, "counter": 9 }, { "number": 224, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": 100, "counter": 8 }, { "number": 211, "counter": 8 }, { "number": 239, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -8, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": -9, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -2, "counter": 7 }, { "number": 36, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": 225, "counter": 6 }, { "number": 282, "counter": 6 }, { "number": -4, "counter": 6 }, { "number": -3, "counter": 6 }, { "number": -10, "counter": 6 }, { "number": 0, "counter": 5 } ], "valid": 537, "invalid": 105, "ratioPresent": 0.918030825942916, "nbUnion": 35385, "duplicates": 176, "invented": 116 }, "meta": { "withName": 635, "withoutName": 7, "withMolfile": 482 }, "info": { "protonNMR": { "present": 597, "notPresent": 45, "valid": 236, "notValid": 361 }, "carbonNMR": { "present": 430, "notPresent": 212, "valid": 166, "notValid": 264 }, "elementalAnalysis": { "present": 264, "notPresent": 378, "valid": 216, "notValid": 48 }, "hrms": { "present": 68, "notPresent": 574, "valid": 11, "notValid": 57 } }, "logs": { "info": 80, "warn": 202, "error": 200, "fatal": 160 }, "chatgpt": { "promptTokens": 519751, "completionTokens": 448254, "totalTokens": 968005, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "cost": 1.4162590000000002 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_1.5_0.5", "version": "v1" } ], "versions": [ "v1" ], "stats": { "yaml": { "parsable": [ 638 ], "notParsable": [ 4 ], "empty": [ 1930 ] }, "numbers": { "valid": [ 537 ], "invalid": [ 105 ], "ratioPresent": [ 0.918030825942916 ], "nbUnion": [ 35385 ], "duplicates": [ 176 ], "invented": [ 116 ] }, "meta": { "withName": [ 635 ], "withoutName": [ 7 ], "withMolfile": [ 482 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_1.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1.5-0.5_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1.5-0.5_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 566, "frequency": { "50": 4, "80": 2, "200": 2, "300": 11, "400": 44, "500": 7, "N/A": 1 }, "solvent": { "unknown": 93, "CDCl3": 95, "not specified": 11, "DMSO-d6": 74, "CD3OD": 32, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 6, "Not specified": 2, "MeOH": 6, "ppm": 2, "MeOD": 3, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "N/A": 1, "acetone-d6": 3, "δ (ppm)": 1 } }, "carbon": { "nbSpectra": 478, "frequency": { "50": 2, "75": 5, "80": 2, "100": 43, "125": 3, "50.3": 1, "75.4": 1, "N/A": 1 }, "solvent": { "unknown": 80, "DMSO-d6": 55, "CDCl3": 89, "not specified": 10, "CD3OD": 30, "CHCl3": 1, "CD3COCD3": 1, "pyridine-d5": 6, "Not specified": 2, "MeOH": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "N/A": 1, "acetone-d6": 5, "δ (ppm)": 1 } } }, "yaml": { "parsable": 625, "notParsable": 17, "empty": 15680 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 169 }, { "number": 400, "counter": 29 }, { "number": 100, "counter": 11 }, { "number": -18, "counter": 2 }, { "number": 20, "counter": 1 }, { "number": 26, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 37, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 142 }, { "number": 1, "counter": 72 }, { "number": 14, "counter": 52 }, { "number": 4, "counter": 48 }, { "number": 100, "counter": 46 }, { "number": 5, "counter": 44 }, { "number": 8, "counter": 38 }, { "number": 10, "counter": 37 }, { "number": 18, "counter": 37 }, { "number": 15, "counter": 36 }, { "number": 17, "counter": 36 }, { "number": 12, "counter": 35 }, { "number": 2, "counter": 34 }, { "number": 9, "counter": 34 }, { "number": 3, "counter": 33 }, { "number": 6, "counter": 33 }, { "number": 11, "counter": 28 }, { "number": 16, "counter": 27 }, { "number": 24, "counter": 23 }, { "number": 25, "counter": 22 }, { "number": 7, "counter": 21 }, { "number": 28, "counter": 19 }, { "number": 20, "counter": 18 }, { "number": 19, "counter": 17 }, { "number": 23, "counter": 17 }, { "number": 224, "counter": 17 }, { "number": 211, "counter": 16 }, { "number": 239, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 22, "counter": 15 }, { "number": 43, "counter": 14 }, { "number": 196, "counter": 14 }, { "number": -2, "counter": 14 }, { "number": 27, "counter": 12 }, { "number": 30, "counter": 12 }, { "number": 254, "counter": 12 }, { "number": -4, "counter": 12 }, { "number": -3, "counter": 12 }, { "number": 135, "counter": 11 }, { "number": -5, "counter": 11 }, { "number": 31, "counter": 10 }, { "number": 70, "counter": 10 }, { "number": 282, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 29, "counter": 9 }, { "number": 32, "counter": 9 }, { "number": 237, "counter": 9 }, { "number": -10, "counter": 9 } ], "valid": 372, "invalid": 270, "ratioPresent": 0.8933193369558714, "nbUnion": 35517, "duplicates": 288, "invented": 614 }, "meta": { "withName": 618, "withoutName": 24, "withMolfile": 470 }, "info": { "protonNMR": { "present": 588, "notPresent": 54, "valid": 170, "notValid": 418 }, "carbonNMR": { "present": 521, "notPresent": 121, "valid": 156, "notValid": 365 }, "elementalAnalysis": { "present": 269, "notPresent": 373, "valid": 223, "notValid": 46 }, "hrms": { "present": 54, "notPresent": 588, "valid": 9, "notValid": 45 } }, "logs": { "info": 37, "warn": 181, "error": 252, "fatal": 172 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 530695, "totalTokens": 1017704, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "cost": 1.548399 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_1.5_0.5", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 625 ], "notParsable": [ 17 ], "empty": [ 15680 ] }, "numbers": { "valid": [ 372 ], "invalid": [ 270 ], "ratioPresent": [ 0.8933193369558714 ], "nbUnion": [ 35517 ], "duplicates": [ 288 ], "invented": [ 614 ] }, "meta": { "withName": [ 618 ], "withoutName": [ 24 ], "withMolfile": [ 470 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_1.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1.5-0.5_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1.5-0.5_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 581, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 29, "500": 4, "N/A": 1, "4.07": 1, "400\"": 1 }, "solvent": { "methanol": 1, "DMSO-d6": 73, "unknown": 6, "N/A": 6, "CDCl3": 89, "CD3OD": 32, "δ/ppm": 1, "not specified": 6, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 5, "CDC13": 1, "pyridine-d5": 7, "literature": 1, "MeOH": 8, "MeOD": 3, "ppm": 1, "CCl4": 1, "D2O": 1, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 2, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 436, "frequency": { "50": 2, "75": 5, "80": 2, "100": 26, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "unknown": 3, "N/A": 4, "CDCl3": 75, "CD3OD": 31, "CHCl3": 1, "CD3COCD3": 1, "pyridine-d5": 7, "MeOH": 3, "CD3CN": 3, "MeOD": 3, "DMSO": 2, "D2O": 1, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "not specified": 2 } } }, "yaml": { "parsable": 627, "notParsable": 15, "empty": 2793 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 16 }, { "number": 100, "counter": 15 }, { "number": 400, "counter": 13 }, { "number": 50, "counter": 3 }, { "number": -18, "counter": 2 }, { "number": 32, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 346, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 5.52, "counter": 1 }, { "number": 5.56, "counter": 1 }, { "number": 20.18, "counter": 1 }, { "number": 20.2, "counter": 1 }, { "number": 57.33, "counter": 1 }, { "number": 57.47, "counter": 1 }, { "number": 391.1, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 441 }, { "number": -1, "counter": 161 }, { "number": 1, "counter": 81 }, { "number": 14, "counter": 59 }, { "number": 10, "counter": 49 }, { "number": 15, "counter": 47 }, { "number": 16, "counter": 42 }, { "number": 5, "counter": 41 }, { "number": 3, "counter": 40 }, { "number": 18, "counter": 40 }, { "number": 2, "counter": 38 }, { "number": 4, "counter": 36 }, { "number": 8, "counter": 36 }, { "number": 6, "counter": 34 }, { "number": 9, "counter": 32 }, { "number": 12, "counter": 32 }, { "number": 17, "counter": 32 }, { "number": 24, "counter": 28 }, { "number": 7, "counter": 24 }, { "number": 11, "counter": 24 }, { "number": 20, "counter": 23 }, { "number": 23, "counter": 22 }, { "number": 25, "counter": 22 }, { "number": 19, "counter": 21 }, { "number": 21, "counter": 19 }, { "number": 22, "counter": 18 }, { "number": 100, "counter": 18 }, { "number": -10, "counter": 16 }, { "number": 27, "counter": 13 }, { "number": -7, "counter": 13 }, { "number": -13, "counter": 12 }, { "number": -8, "counter": 12 }, { "number": 28, "counter": 11 }, { "number": -3, "counter": 10 }, { "number": 30, "counter": 9 }, { "number": 31, "counter": 9 }, { "number": -4, "counter": 8 }, { "number": -12, "counter": 8 }, { "number": -9, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 33, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": -2, "counter": 7 }, { "number": 225, "counter": 6 }, { "number": 7.5, "counter": 6 }, { "number": 8.1, "counter": 6 }, { "number": -6, "counter": 6 }, { "number": 0, "counter": 5 }, { "number": 29, "counter": 5 }, { "number": 209, "counter": 5 } ], "valid": 508, "invalid": 134, "ratioPresent": 0.9041893830959603, "nbUnion": 35360, "duplicates": 302, "invented": 98 }, "meta": { "withName": 617, "withoutName": 25, "withMolfile": 464 }, "info": { "protonNMR": { "present": 584, "notPresent": 58, "valid": 223, "notValid": 361 }, "carbonNMR": { "present": 439, "notPresent": 203, "valid": 174, "notValid": 265 }, "elementalAnalysis": { "present": 254, "notPresent": 388, "valid": 214, "notValid": 40 }, "hrms": { "present": 82, "notPresent": 560, "valid": 14, "notValid": 68 } }, "logs": { "info": 66, "warn": 216, "error": 181, "fatal": 179 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 469305, "totalTokens": 946042, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "cost": 1.4153470000000001 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_1.5_0.5", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 627 ], "notParsable": [ 15 ], "empty": [ 2793 ] }, "numbers": { "valid": [ 508 ], "invalid": [ 134 ], "ratioPresent": [ 0.9041893830959603 ], "nbUnion": [ 35360 ], "duplicates": [ 302 ], "invented": [ 98 ] }, "meta": { "withName": [ 617 ], "withoutName": [ 25 ], "withMolfile": [ 464 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_1.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1.5-0.5_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1.5-0.5_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 566, "frequency": { "50": 4, "80": 2, "200": 3, "300": 9, "400": 45, "500": 3, "not specified": 1 }, "solvent": { "CDCl3": 90, "not specified": 41, "CD3OD": 32, "Not specified": 11, "CHCl3": 2, "DMSO-d6": 68, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 9, "CDC13": 1, "unknown": 4, "pyridine-d5": 7, "not provided": 2, "MeOD": 3, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 446, "frequency": { "50": 2, "75": 5, "80": 2, "100": 42, "125": 3, "200": 1, "400": 1, "not specified": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CDCl3": 76, "not specified": 32, "CD3OD": 31, "Not specified": 9, "CHCl3": 2, "MeOH": 8, "CD3COCD3": 1, "unknown": 4, "pyridine-d5": 7, "not provided": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "CH3OH": 1 } } }, "yaml": { "parsable": 602, "notParsable": 40, "empty": 2159 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 45 }, { "number": 400, "counter": 28 }, { "number": 0, "counter": 12 }, { "number": 50, "counter": 7 }, { "number": 15, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": -3, "counter": 2 }, { "number": 28, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": -18.7, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 437 }, { "number": -1, "counter": 190 }, { "number": 1, "counter": 139 }, { "number": 3, "counter": 70 }, { "number": 2, "counter": 63 }, { "number": 6, "counter": 59 }, { "number": 4, "counter": 56 }, { "number": 5, "counter": 53 }, { "number": 8, "counter": 50 }, { "number": 14, "counter": 39 }, { "number": 7, "counter": 36 }, { "number": 10, "counter": 35 }, { "number": 100, "counter": 33 }, { "number": 9, "counter": 30 }, { "number": 12, "counter": 30 }, { "number": 15, "counter": 30 }, { "number": 11, "counter": 29 }, { "number": 18, "counter": 25 }, { "number": 16, "counter": 23 }, { "number": 17, "counter": 23 }, { "number": 24, "counter": 21 }, { "number": 25, "counter": 21 }, { "number": -3, "counter": 21 }, { "number": -2, "counter": 20 }, { "number": -4, "counter": 20 }, { "number": 23, "counter": 19 }, { "number": 19, "counter": 16 }, { "number": 28, "counter": 15 }, { "number": 21, "counter": 13 }, { "number": 224, "counter": 13 }, { "number": -5, "counter": 12 }, { "number": 20, "counter": 11 }, { "number": 50, "counter": 11 }, { "number": 63, "counter": 11 }, { "number": 211, "counter": 11 }, { "number": 22, "counter": 10 }, { "number": 239, "counter": 10 }, { "number": 90, "counter": 9 }, { "number": 196, "counter": 9 }, { "number": 237, "counter": 9 }, { "number": 310, "counter": 9 }, { "number": -7, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": 70, "counter": 8 }, { "number": 254, "counter": 8 }, { "number": 267, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 0, "counter": 7 }, { "number": 26, "counter": 7 }, { "number": 282, "counter": 7 } ], "valid": 483, "invalid": 159, "ratioPresent": 0.8736272760198069, "nbUnion": 35407, "duplicates": 250, "invented": 155 }, "meta": { "withName": 589, "withoutName": 53, "withMolfile": 450 }, "info": { "protonNMR": { "present": 567, "notPresent": 75, "valid": 201, "notValid": 366 }, "carbonNMR": { "present": 446, "notPresent": 196, "valid": 155, "notValid": 291 }, "elementalAnalysis": { "present": 234, "notPresent": 408, "valid": 194, "notValid": 40 }, "hrms": { "present": 95, "notPresent": 547, "valid": 11, "notValid": 84 } }, "logs": { "info": 64, "warn": 150, "error": 236, "fatal": 192 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 464510, "totalTokens": 938679, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "cost": 1.403189 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_1.5_0.5", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 602 ], "notParsable": [ 40 ], "empty": [ 2159 ] }, "numbers": { "valid": [ 483 ], "invalid": [ 159 ], "ratioPresent": [ 0.8736272760198069 ], "nbUnion": [ 35407 ], "duplicates": [ 250 ], "invented": [ 155 ] }, "meta": { "withName": [ 589 ], "withoutName": [ 53 ], "withMolfile": [ 450 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_1.5_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-1.5-0.5_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-1.5-0.5_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 590, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 20, "500": 6 }, "solvent": { "CDCl3": 87, "unknown": 1, "DMSO-d6": 74, "CD3OD": 31, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "MeOH": 8, "CDC13": 1, "pyridine-d5": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 434, "frequency": { "50": 2, "75": 5, "80": 2, "100": 18, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 51, "CD3OD": 28, "CHCl3": 2, "CDCl3": 69, "CD3COCD3": 1, "pyridine-d5": 7, "MeOH": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 642, "notParsable": 0, "empty": 1984 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 9 }, { "number": 100, "counter": 8 }, { "number": 50, "counter": 3 }, { "number": 400, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 15, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 179 }, { "number": 1, "counter": 76 }, { "number": 14, "counter": 56 }, { "number": 10, "counter": 46 }, { "number": 16, "counter": 44 }, { "number": 4, "counter": 42 }, { "number": 3, "counter": 41 }, { "number": 18, "counter": 39 }, { "number": 6, "counter": 37 }, { "number": 5, "counter": 36 }, { "number": 8, "counter": 36 }, { "number": 9, "counter": 36 }, { "number": 15, "counter": 35 }, { "number": 17, "counter": 32 }, { "number": 12, "counter": 28 }, { "number": 2, "counter": 25 }, { "number": 24, "counter": 25 }, { "number": -10, "counter": 24 }, { "number": 11, "counter": 23 }, { "number": 23, "counter": 23 }, { "number": 25, "counter": 23 }, { "number": 19, "counter": 21 }, { "number": 7, "counter": 19 }, { "number": 21, "counter": 18 }, { "number": -8, "counter": 18 }, { "number": -7, "counter": 18 }, { "number": 28, "counter": 16 }, { "number": -9, "counter": 16 }, { "number": 20, "counter": 14 }, { "number": 22, "counter": 14 }, { "number": 100, "counter": 12 }, { "number": -13, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 27, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 224, "counter": 8 }, { "number": -12, "counter": 8 }, { "number": 32, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 211, "counter": 7 }, { "number": 239, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -3, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 196, "counter": 6 }, { "number": 209, "counter": 6 }, { "number": 240, "counter": 6 } ], "valid": 523, "invalid": 119, "ratioPresent": 0.9158588721003571, "nbUnion": 35331, "duplicates": 265, "invented": 60 }, "meta": { "withName": 618, "withoutName": 24, "withMolfile": 466 }, "info": { "protonNMR": { "present": 590, "notPresent": 52, "valid": 233, "notValid": 357 }, "carbonNMR": { "present": 434, "notPresent": 208, "valid": 173, "notValid": 261 }, "elementalAnalysis": { "present": 255, "notPresent": 387, "valid": 217, "notValid": 38 }, "hrms": { "present": 86, "notPresent": 556, "valid": 12, "notValid": 74 } }, "logs": { "info": 69, "warn": 213, "error": 184, "fatal": 176 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 441384, "totalTokens": 923899, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 1.5, "top_p": 0.5, "cost": 1.365283 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_1.5_0.5", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 642 ], "notParsable": [ 0 ], "empty": [ 1984 ] }, "numbers": { "valid": [ 523 ], "invalid": [ 119 ], "ratioPresent": [ 0.9158588721003571 ], "nbUnion": [ 35331 ], "duplicates": [ 265 ], "invented": [ 60 ] }, "meta": { "withName": [ 618 ], "withoutName": [ 24 ], "withMolfile": [ 466 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-3.5-turbo-1106_2_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-2-0.5_v1.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-2-0.5_v1" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 582, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 20, "500": 4 }, "solvent": { "CDCl3": 84, "not specified": 2, "DMSO-d6": 74, "unknown": 1, "CD3OD": 32, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 425, "frequency": { "50": 2, "75": 5, "80": 2, "100": 16, "125": 3, "200": 2, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "unknown": 1, "CD3OD": 29, "CHCl3": 1, "CDCl3": 66, "CD3COCD3": 1, "pyridine-d5": 6, "MeOH": 6, "not specified": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 635, "notParsable": 7, "empty": 1254 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 13 }, { "number": 0, "counter": 6 }, { "number": 50, "counter": 3 }, { "number": 400, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 11, "counter": 1 }, { "number": 14, "counter": 1 }, { "number": 25, "counter": 1 }, { "number": 32, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 432 }, { "number": -1, "counter": 187 }, { "number": 1, "counter": 85 }, { "number": 14, "counter": 48 }, { "number": 4, "counter": 45 }, { "number": 10, "counter": 45 }, { "number": 18, "counter": 43 }, { "number": 3, "counter": 42 }, { "number": 6, "counter": 42 }, { "number": 16, "counter": 40 }, { "number": 2, "counter": 38 }, { "number": 5, "counter": 38 }, { "number": 8, "counter": 38 }, { "number": 9, "counter": 36 }, { "number": 15, "counter": 35 }, { "number": 17, "counter": 35 }, { "number": 12, "counter": 30 }, { "number": 24, "counter": 26 }, { "number": 11, "counter": 25 }, { "number": -10, "counter": 25 }, { "number": 20, "counter": 21 }, { "number": 25, "counter": 21 }, { "number": -8, "counter": 21 }, { "number": 7, "counter": 19 }, { "number": -7, "counter": 19 }, { "number": 19, "counter": 18 }, { "number": 23, "counter": 17 }, { "number": 21, "counter": 16 }, { "number": 22, "counter": 16 }, { "number": 100, "counter": 15 }, { "number": -2, "counter": 15 }, { "number": -9, "counter": 15 }, { "number": -13, "counter": 15 }, { "number": 28, "counter": 13 }, { "number": -3, "counter": 12 }, { "number": -4, "counter": 10 }, { "number": 70, "counter": 9 }, { "number": -12, "counter": 9 }, { "number": 30, "counter": 8 }, { "number": 224, "counter": 8 }, { "number": 27, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 26, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": 211, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": 267, "counter": 6 }, { "number": 4.2, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": 29, "counter": 5 } ], "valid": 555, "invalid": 87, "ratioPresent": 0.9136269274376281, "nbUnion": 35334, "duplicates": 152, "invented": 63 }, "meta": { "withName": 615, "withoutName": 27, "withMolfile": 465 }, "info": { "protonNMR": { "present": 584, "notPresent": 58, "valid": 224, "notValid": 360 }, "carbonNMR": { "present": 426, "notPresent": 216, "valid": 176, "notValid": 250 }, "elementalAnalysis": { "present": 252, "notPresent": 390, "valid": 213, "notValid": 39 }, "hrms": { "present": 78, "notPresent": 564, "valid": 14, "notValid": 64 } }, "logs": { "info": 73, "warn": 204, "error": 188, "fatal": 177 }, "chatgpt": { "promptTokens": 488935, "completionTokens": 439869, "totalTokens": 928804, "nbPrompts": 642, "version": "v1", "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "cost": 1.368673 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-3.5-turbo-1106_2_0.5", "version": "v1" } ], "versions": [ "v1" ], "stats": { "yaml": { "parsable": [ 635 ], "notParsable": [ 7 ], "empty": [ 1254 ] }, "numbers": { "valid": [ 555 ], "invalid": [ 87 ], "ratioPresent": [ 0.9136269274376281 ], "nbUnion": [ 35334 ], "duplicates": [ 152 ], "invented": [ 63 ] }, "meta": { "withName": [ 615 ], "withoutName": [ 27 ], "withMolfile": [ 465 ] } }, "description": "Our baseline prompt" }, { "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_2_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-2-0.5_v2.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-2-0.5_v2" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Testing if question mark is really useful" }, "stats": { "nmr": { "proton": { "nbSpectra": 562, "frequency": { "50": 4, "80": 2, "200": 3, "300": 11, "400": 38, "500": 7, "1.21": 1 }, "solvent": { "CDCl3": 93, "not specified": 19, "unknown": 94, "DMSO-d6": 74, "CD3OD": 28, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "N/A": 2, "literature": 1, "pyridine-d5": 7, "MeOH": 6, "ppm": 1, "MeOD": 3, "dimethylformamide": 1, "δ/ppm": 1, "D2O": 4, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 475, "frequency": { "50": 2, "75": 5, "80": 2, "100": 37, "125": 3, "200": 1, "400": 1, "50.3": 1, "75.4": 1 }, "solvent": { "unknown": 78, "DMSO-d6": 57, "not specified": 18, "CDCl3": 88, "CD3OD": 26, "CHCl3": 1, "CD3COCD3": 1, "N/A": 2, "literature": 1, "pyridine-d5": 7, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "MeOH": 4, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 625, "notParsable": 17, "empty": 15320 }, "numbers": { "inventedNumbers": [ { "number": 0, "counter": 161 }, { "number": 400, "counter": 20 }, { "number": 100, "counter": 8 }, { "number": 29, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 3, "counter": 1 }, { "number": 20, "counter": 1 }, { "number": 27, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 39, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": -1700, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 434 }, { "number": -1, "counter": 141 }, { "number": 1, "counter": 78 }, { "number": 14, "counter": 58 }, { "number": 100, "counter": 45 }, { "number": 4, "counter": 42 }, { "number": 15, "counter": 41 }, { "number": 2, "counter": 39 }, { "number": 5, "counter": 39 }, { "number": 10, "counter": 39 }, { "number": 18, "counter": 39 }, { "number": 8, "counter": 38 }, { "number": 3, "counter": 36 }, { "number": 9, "counter": 35 }, { "number": 6, "counter": 33 }, { "number": 16, "counter": 32 }, { "number": 11, "counter": 30 }, { "number": 12, "counter": 30 }, { "number": 17, "counter": 29 }, { "number": 7, "counter": 25 }, { "number": 24, "counter": 25 }, { "number": 25, "counter": 25 }, { "number": 19, "counter": 22 }, { "number": 20, "counter": 21 }, { "number": 21, "counter": 18 }, { "number": 23, "counter": 18 }, { "number": 224, "counter": 18 }, { "number": 22, "counter": 17 }, { "number": 28, "counter": 16 }, { "number": 239, "counter": 16 }, { "number": 267, "counter": 16 }, { "number": -3, "counter": 16 }, { "number": -2, "counter": 16 }, { "number": 211, "counter": 15 }, { "number": -10, "counter": 15 }, { "number": -8, "counter": 15 }, { "number": 196, "counter": 14 }, { "number": 27, "counter": 13 }, { "number": 30, "counter": 13 }, { "number": -13, "counter": 13 }, { "number": -4, "counter": 13 }, { "number": -9, "counter": 13 }, { "number": 26, "counter": 12 }, { "number": 43, "counter": 12 }, { "number": 135, "counter": 11 }, { "number": 254, "counter": 11 }, { "number": 282, "counter": 11 }, { "number": -14, "counter": 11 }, { "number": -5, "counter": 11 }, { "number": 70, "counter": 10 } ], "valid": 363, "invalid": 279, "ratioPresent": 0.8881962731247118, "nbUnion": 35499, "duplicates": 386, "invented": 614 }, "meta": { "withName": 619, "withoutName": 23, "withMolfile": 477 }, "info": { "protonNMR": { "present": 580, "notPresent": 62, "valid": 160, "notValid": 420 }, "carbonNMR": { "present": 515, "notPresent": 127, "valid": 152, "notValid": 363 }, "elementalAnalysis": { "present": 274, "notPresent": 368, "valid": 225, "notValid": 49 }, "hrms": { "present": 56, "notPresent": 586, "valid": 8, "notValid": 48 } }, "logs": { "info": 35, "warn": 185, "error": 257, "fatal": 165 }, "chatgpt": { "promptTokens": 487009, "completionTokens": 527633, "totalTokens": 1014642, "nbPrompts": 642, "version": "v2", "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "cost": 1.542275 } }, "description": "Testing if question mark is really useful", "key": "Testing if question mark is really useful_gpt-3.5-turbo-1106_2_0.5", "version": "v2" } ], "versions": [ "v2" ], "stats": { "yaml": { "parsable": [ 625 ], "notParsable": [ 17 ], "empty": [ 15320 ] }, "numbers": { "valid": [ 363 ], "invalid": [ 279 ], "ratioPresent": [ 0.8881962731247118 ], "nbUnion": [ 35499 ], "duplicates": [ 386 ], "invented": [ 614 ] }, "meta": { "withName": [ 619 ], "withoutName": [ 23 ], "withMolfile": [ 477 ] } }, "description": "Testing if question mark is really useful" }, { "key": "Remove Do not guess_gpt-3.5-turbo-1106_2_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-2-0.5_v3.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-2-0.5_v3" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Remove Do not guess" }, "stats": { "nmr": { "proton": { "nbSpectra": 583, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 28, "500": 5 }, "solvent": { "CDCl3": 83, "methanol": 1, "N/A": 6, "DMSO-d6": 72, "unknown": 9, "Et2O": 1, "not specified": 4, "CD3OD": 33, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "Not specified": 2, "pyridine-d5": 7, "MeOH": 9, "MeOD": 3, "dimethylformamide": 1, "D2O": 1, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "d": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 437, "frequency": { "50": 2, "75": 5, "80": 2, "100": 22, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 50, "unknown": 5, "N/A": 3, "Et2O": 1, "not specified": 2, "CD3OD": 32, "CHCl3": 1, "CDCl3": 69, "CD3COCD3": 1, "Not specified": 2, "pyridine-d5": 7, "MeOH": 6, "CD3CN": 4, "MeOD": 3, "DMSO": 1, "unspecified": 1, "dimethylformamide": 1, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "CH3OH": 1 } } }, "yaml": { "parsable": 627, "notParsable": 15, "empty": 2880 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 14 }, { "number": 0, "counter": 13 }, { "number": 400, "counter": 10 }, { "number": 29, "counter": 2 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 1, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 18, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 33, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": -3, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 438 }, { "number": -1, "counter": 178 }, { "number": 1, "counter": 76 }, { "number": 14, "counter": 63 }, { "number": 3, "counter": 47 }, { "number": 10, "counter": 45 }, { "number": 15, "counter": 45 }, { "number": 16, "counter": 44 }, { "number": 18, "counter": 44 }, { "number": 5, "counter": 39 }, { "number": 4, "counter": 38 }, { "number": 6, "counter": 38 }, { "number": 12, "counter": 38 }, { "number": 2, "counter": 37 }, { "number": 8, "counter": 35 }, { "number": 17, "counter": 35 }, { "number": 9, "counter": 30 }, { "number": 20, "counter": 25 }, { "number": 24, "counter": 24 }, { "number": -10, "counter": 24 }, { "number": 7, "counter": 22 }, { "number": -8, "counter": 22 }, { "number": 11, "counter": 21 }, { "number": 19, "counter": 21 }, { "number": 25, "counter": 21 }, { "number": -7, "counter": 21 }, { "number": 23, "counter": 20 }, { "number": 21, "counter": 18 }, { "number": -13, "counter": 17 }, { "number": 22, "counter": 15 }, { "number": 28, "counter": 15 }, { "number": 100, "counter": 15 }, { "number": 26, "counter": 11 }, { "number": -2, "counter": 11 }, { "number": -9, "counter": 11 }, { "number": -12, "counter": 11 }, { "number": -3, "counter": 10 }, { "number": 27, "counter": 9 }, { "number": -6, "counter": 8 }, { "number": 30, "counter": 7 }, { "number": 224, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": -5, "counter": 7 }, { "number": -4, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 32, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 43, "counter": 6 }, { "number": 70, "counter": 6 }, { "number": 225, "counter": 6 } ], "valid": 502, "invalid": 140, "ratioPresent": 0.9041525550954007, "nbUnion": 35351, "duplicates": 294, "invented": 91 }, "meta": { "withName": 612, "withoutName": 30, "withMolfile": 460 }, "info": { "protonNMR": { "present": 585, "notPresent": 57, "valid": 233, "notValid": 352 }, "carbonNMR": { "present": 438, "notPresent": 204, "valid": 164, "notValid": 274 }, "elementalAnalysis": { "present": 257, "notPresent": 385, "valid": 214, "notValid": 43 }, "hrms": { "present": 76, "notPresent": 566, "valid": 13, "notValid": 63 } }, "logs": { "info": 64, "warn": 219, "error": 177, "fatal": 182 }, "chatgpt": { "promptTokens": 476737, "completionTokens": 468499, "totalTokens": 945236, "nbPrompts": 642, "version": "v3", "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "cost": 1.4137350000000002 } }, "description": "Remove Do not guess", "key": "Remove Do not guess_gpt-3.5-turbo-1106_2_0.5", "version": "v3" } ], "versions": [ "v3" ], "stats": { "yaml": { "parsable": [ 627 ], "notParsable": [ 15 ], "empty": [ 2880 ] }, "numbers": { "valid": [ 502 ], "invalid": [ 140 ], "ratioPresent": [ 0.9041525550954007 ], "nbUnion": [ 35351 ], "duplicates": [ 294 ], "invented": [ 91 ] }, "meta": { "withName": [ 612 ], "withoutName": [ 30 ], "withMolfile": [ 460 ] } }, "description": "Remove Do not guess" }, { "key": "Is context useful?_gpt-3.5-turbo-1106_2_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Is context useful?" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-2-0.5_v4.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-2-0.5_v4" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Is context useful?" }, "stats": { "nmr": { "proton": { "nbSpectra": 565, "frequency": { "20": 1, "50": 4, "80": 2, "200": 4, "300": 10, "400": 35, "500": 3, "Not specified": 2 }, "solvent": { "DMSO-d6": 70, "not specified": 34, "CDCl3": 88, "CD3OD": 32, "unknown": 4, "Not specified": 10, "CHCl3": 2, "CD3COCD3": 2, "CD3CN": 5, "MeOH": 10, "CDC13": 1, "pyridine-d5": 7, "MeOD": 3, "D2O": 2, "d4-MeOH": 36, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "not provided": 1, "δ": 1, "CH3OH": 1 } }, "carbon": { "nbSpectra": 438, "frequency": { "20": 1, "50": 2, "75": 4, "80": 2, "100": 32, "125": 3, "200": 2, "400": 1, "not provided": 1, "75.5": 1, "50.3": 1, "75.4": 1, "20.6": 1, "100.6": 1 }, "solvent": { "DMSO-d6": 50, "not specified": 25, "CDCl3": 77, "CD3OD": 31, "not provided": 2, "CHCl3": 2, "MeOH": 9, "unknown": 3, "CD3COCD3": 1, "Not specified": 5, "pyridine-d5": 7, "CD3CN": 3, "MeOD": 3, "DMSO": 1, "qz": 1, "D2O": 2, "d4-MeOH": 34, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "δ": 1, "CH3OH": 1 } } }, "yaml": { "parsable": 605, "notParsable": 37, "empty": 2087 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 43 }, { "number": 400, "counter": 20 }, { "number": 0, "counter": 9 }, { "number": 50, "counter": 8 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 6, "counter": 1 }, { "number": 7, "counter": 1 }, { "number": 8, "counter": 1 }, { "number": 9, "counter": 1 }, { "number": 10, "counter": 1 }, { "number": 11, "counter": 1 }, { "number": 15, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 75.5, "counter": 1 }, { "number": 296.5, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 100.6, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": -3, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 430 }, { "number": -1, "counter": 186 }, { "number": 1, "counter": 140 }, { "number": 2, "counter": 67 }, { "number": 3, "counter": 66 }, { "number": 6, "counter": 59 }, { "number": 8, "counter": 57 }, { "number": 4, "counter": 54 }, { "number": 5, "counter": 49 }, { "number": 10, "counter": 47 }, { "number": 14, "counter": 41 }, { "number": 9, "counter": 38 }, { "number": 100, "counter": 33 }, { "number": 15, "counter": 32 }, { "number": 7, "counter": 31 }, { "number": 12, "counter": 29 }, { "number": 18, "counter": 28 }, { "number": 11, "counter": 27 }, { "number": 17, "counter": 26 }, { "number": 24, "counter": 24 }, { "number": 16, "counter": 23 }, { "number": -3, "counter": 23 }, { "number": 25, "counter": 22 }, { "number": -2, "counter": 22 }, { "number": 23, "counter": 15 }, { "number": 19, "counter": 14 }, { "number": -4, "counter": 14 }, { "number": 20, "counter": 11 }, { "number": 21, "counter": 11 }, { "number": 22, "counter": 11 }, { "number": 28, "counter": 11 }, { "number": 63, "counter": 9 }, { "number": 70, "counter": 9 }, { "number": 90, "counter": 9 }, { "number": 211, "counter": 9 }, { "number": -5, "counter": 9 }, { "number": 27, "counter": 8 }, { "number": 50, "counter": 8 }, { "number": 224, "counter": 8 }, { "number": 225, "counter": 8 }, { "number": 26, "counter": 7 }, { "number": 83, "counter": 7 }, { "number": 196, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -8, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 31, "counter": 6 }, { "number": 239, "counter": 6 }, { "number": 254, "counter": 6 }, { "number": 287, "counter": 6 } ], "valid": 483, "invalid": 159, "ratioPresent": 0.878411076705524, "nbUnion": 35403, "duplicates": 234, "invented": 140 }, "meta": { "withName": 593, "withoutName": 49, "withMolfile": 443 }, "info": { "protonNMR": { "present": 566, "notPresent": 76, "valid": 200, "notValid": 366 }, "carbonNMR": { "present": 438, "notPresent": 204, "valid": 163, "notValid": 275 }, "elementalAnalysis": { "present": 232, "notPresent": 410, "valid": 198, "notValid": 34 }, "hrms": { "present": 97, "notPresent": 545, "valid": 11, "notValid": 86 } }, "logs": { "info": 62, "warn": 150, "error": 231, "fatal": 199 }, "chatgpt": { "promptTokens": 474169, "completionTokens": 461220, "totalTokens": 935389, "nbPrompts": 642, "version": "v4", "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "cost": 1.396609 } }, "description": "Is context useful?", "key": "Is context useful?_gpt-3.5-turbo-1106_2_0.5", "version": "v4" } ], "versions": [ "v4" ], "stats": { "yaml": { "parsable": [ 605 ], "notParsable": [ 37 ], "empty": [ 2087 ] }, "numbers": { "valid": [ 483 ], "invalid": [ 159 ], "ratioPresent": [ 0.878411076705524 ], "nbUnion": [ 35403 ], "duplicates": [ 234 ], "invented": [ 140 ] }, "meta": { "withName": [ 593 ], "withoutName": [ 49 ], "withMolfile": [ 443 ] } }, "description": "Is context useful?" }, { "key": "Be positive_gpt-3.5-turbo-1106_2_0.5", "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Be positive" }, "entries": [ { "file": { "name": "gpt-3.5-turbo-1106-2-0.5_v5.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-3.5-turbo-1106-2-0.5_v5" }, "model": { "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "promptDescription": "Be positive" }, "stats": { "nmr": { "proton": { "nbSpectra": 588, "frequency": { "50": 4, "80": 2, "200": 4, "300": 11, "400": 22, "500": 4 }, "solvent": { "CDCl3": 86, "unknown": 1, "DMSO-d6": 73, "CD3OD": 32, "CHCl3": 1, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "N/A": 1, "pyridine-d5": 7, "MeOH": 7, "MeOD": 3, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "acetone-d6": 3, "CH3OH": 1 } }, "carbon": { "nbSpectra": 430, "frequency": { "50": 2, "75": 5, "80": 2, "100": 18, "125": 3, "200": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 49, "CD3OD": 30, "CHCl3": 1, "CDCl3": 69, "CD3COCD3": 1, "N/A": 1, "pyridine-d5": 7, "MeOH": 5, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "acetone-d6": 5, "CH3OH": 1 } } }, "yaml": { "parsable": 639, "notParsable": 3, "empty": 1812 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 9 }, { "number": 0, "counter": 7 }, { "number": 400, "counter": 4 }, { "number": 50, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 29, "counter": 1 }, { "number": 31, "counter": 1 }, { "number": 34, "counter": 1 }, { "number": 35, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.08, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": -3, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 149.8, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 432 }, { "number": -1, "counter": 180 }, { "number": 1, "counter": 77 }, { "number": 14, "counter": 54 }, { "number": 10, "counter": 47 }, { "number": 15, "counter": 45 }, { "number": 4, "counter": 44 }, { "number": 16, "counter": 44 }, { "number": 8, "counter": 40 }, { "number": 3, "counter": 39 }, { "number": 5, "counter": 39 }, { "number": 18, "counter": 37 }, { "number": 12, "counter": 36 }, { "number": 6, "counter": 34 }, { "number": 9, "counter": 34 }, { "number": 2, "counter": 33 }, { "number": 17, "counter": 30 }, { "number": 24, "counter": 26 }, { "number": 25, "counter": 24 }, { "number": -10, "counter": 24 }, { "number": 23, "counter": 23 }, { "number": 11, "counter": 22 }, { "number": 19, "counter": 21 }, { "number": -7, "counter": 21 }, { "number": 7, "counter": 20 }, { "number": -8, "counter": 20 }, { "number": 20, "counter": 19 }, { "number": 21, "counter": 19 }, { "number": -9, "counter": 17 }, { "number": 100, "counter": 15 }, { "number": 28, "counter": 14 }, { "number": -13, "counter": 13 }, { "number": 22, "counter": 11 }, { "number": 27, "counter": 11 }, { "number": -4, "counter": 11 }, { "number": 26, "counter": 10 }, { "number": 30, "counter": 10 }, { "number": -3, "counter": 10 }, { "number": -5, "counter": 10 }, { "number": -2, "counter": 9 }, { "number": -12, "counter": 9 }, { "number": 224, "counter": 8 }, { "number": -6, "counter": 8 }, { "number": 33, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 211, "counter": 7 }, { "number": 267, "counter": 7 }, { "number": -11, "counter": 7 }, { "number": 239, "counter": 6 }, { "number": 240, "counter": 6 } ], "valid": 533, "invalid": 109, "ratioPresent": 0.910312112081612, "nbUnion": 35332, "duplicates": 277, "invented": 55 }, "meta": { "withName": 618, "withoutName": 24, "withMolfile": 462 }, "info": { "protonNMR": { "present": 588, "notPresent": 54, "valid": 229, "notValid": 359 }, "carbonNMR": { "present": 430, "notPresent": 212, "valid": 168, "notValid": 262 }, "elementalAnalysis": { "present": 247, "notPresent": 395, "valid": 209, "notValid": 38 }, "hrms": { "present": 77, "notPresent": 565, "valid": 11, "notValid": 66 } }, "logs": { "info": 70, "warn": 209, "error": 183, "fatal": 180 }, "chatgpt": { "promptTokens": 482515, "completionTokens": 439591, "totalTokens": 922106, "nbPrompts": 642, "version": "v5", "model": "gpt-3.5-turbo-1106", "temperature": 2, "top_p": 0.5, "cost": 1.361697 } }, "description": "Be positive", "key": "Be positive_gpt-3.5-turbo-1106_2_0.5", "version": "v5" } ], "versions": [ "v5" ], "stats": { "yaml": { "parsable": [ 639 ], "notParsable": [ 3 ], "empty": [ 1812 ] }, "numbers": { "valid": [ 533 ], "invalid": [ 109 ], "ratioPresent": [ 0.910312112081612 ], "nbUnion": [ 35332 ], "duplicates": [ 277 ], "invented": [ 55 ] }, "meta": { "withName": [ 618 ], "withoutName": [ 24 ], "withMolfile": [ 462 ] } }, "description": "Be positive" }, { "key": "Our baseline prompt_gpt-4_0.2_0.2", "model": { "model": "gpt-4", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-4-0.2-0.2_v200.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-4-0.2-0.2_v200" }, "model": { "model": "gpt-4", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 554, "frequency": { "50": 3, "80": 2, "200": 1, "300": 11, "400": 18, "500": 7 }, "solvent": { "CDCl3": 87, "CD3OD": 34, "DMSO-d6": 72, "CD3COCD3": 2, "CD3CN": 6, "DMSO-d6, 50 °C": 1, "CDC13": 1, "pyridine-d5": 7, "See Table 1": 1, "MeOD": 3, "d4-MeOH": 31, "CDCl3 + d4-MeOH": 2, "DMSO": 1, "Table1": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 414, "frequency": { "50": 2, "75": 5, "80": 2, "100": 17, "125": 3, "50.3": 1 }, "solvent": { "CD3OD": 32, "DMSO-d6": 51, "CDCl3": 67, "CD3COCD3": 1, "pyridine-d5": 7, "See Table 1": 1, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 29, "CDCl3 + d4-MeOH": 2, "acetone-d6": 4, "Table1": 1 } } }, "yaml": { "parsable": 594, "notParsable": 48, "empty": 63 }, "numbers": { "inventedNumbers": [ { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 149.8, "counter": 1 }, { "number": 115.4, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 440 }, { "number": -1, "counter": 201 }, { "number": 1, "counter": 78 }, { "number": 4, "counter": 69 }, { "number": 3, "counter": 60 }, { "number": 6, "counter": 57 }, { "number": 5, "counter": 53 }, { "number": 2, "counter": 50 }, { "number": 16, "counter": 50 }, { "number": 14, "counter": 46 }, { "number": 8, "counter": 40 }, { "number": 15, "counter": 39 }, { "number": 18, "counter": 37 }, { "number": 12, "counter": 35 }, { "number": 17, "counter": 34 }, { "number": 10, "counter": 32 }, { "number": 7, "counter": 25 }, { "number": 9, "counter": 25 }, { "number": 24, "counter": 23 }, { "number": -3, "counter": 23 }, { "number": 23, "counter": 22 }, { "number": -2, "counter": 22 }, { "number": 25, "counter": 20 }, { "number": 100, "counter": 20 }, { "number": 11, "counter": 19 }, { "number": 19, "counter": 19 }, { "number": 20, "counter": 19 }, { "number": -5, "counter": 19 }, { "number": -4, "counter": 16 }, { "number": 21, "counter": 15 }, { "number": 22, "counter": 15 }, { "number": 28, "counter": 15 }, { "number": 27, "counter": 13 }, { "number": -8, "counter": 12 }, { "number": -6, "counter": 12 }, { "number": 70, "counter": 11 }, { "number": -7, "counter": 10 }, { "number": 7.3, "counter": 10 }, { "number": 26, "counter": 9 }, { "number": 135, "counter": 9 }, { "number": 7.2, "counter": 8 }, { "number": 8.8, "counter": 8 }, { "number": -11, "counter": 8 }, { "number": 1447, "counter": 7 }, { "number": 6.8, "counter": 7 }, { "number": 128.1, "counter": 7 }, { "number": 30, "counter": 6 }, { "number": 33, "counter": 6 }, { "number": 35, "counter": 6 }, { "number": 77, "counter": 6 } ], "valid": 607, "invalid": 35, "ratioPresent": 0.8672665270092629, "nbUnion": 35314, "duplicates": 17, "invented": 21 }, "meta": { "withName": 593, "withoutName": 49, "withMolfile": 417 }, "info": { "protonNMR": { "present": 555, "notPresent": 87, "valid": 239, "notValid": 316 }, "carbonNMR": { "present": 415, "notPresent": 227, "valid": 167, "notValid": 248 }, "elementalAnalysis": { "present": 233, "notPresent": 409, "valid": 198, "notValid": 35 }, "hrms": { "present": 56, "notPresent": 586, "valid": 14, "notValid": 42 } }, "logs": { "info": 93, "warn": 209, "error": 115, "fatal": 225 }, "chatgpt": { "promptTokens": 488935, "completionTokens": 461074, "totalTokens": 950009, "nbPrompts": 642, "version": "v200", "model": "gpt-4", "temperature": 0.2, "top_p": 0.2, "cost": 42.33249 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-4_0.2_0.2", "version": "v200" } ], "versions": [ "v200" ], "stats": { "yaml": { "parsable": [ 594 ], "notParsable": [ 48 ], "empty": [ 63 ] }, "numbers": { "valid": [ 607 ], "invalid": [ 35 ], "ratioPresent": [ 0.8672665270092629 ], "nbUnion": [ 35314 ], "duplicates": [ 17 ], "invented": [ 21 ] }, "meta": { "withName": [ 593 ], "withoutName": [ 49 ], "withMolfile": [ 417 ] } }, "description": "Our baseline prompt" }, { "key": "Our baseline prompt_gpt-4-1106-preview_0.2_0.2", "model": { "model": "gpt-4-1106-preview", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt" }, "entries": [ { "file": { "name": "gpt-4-1106-preview-0.2-0.2_v200.json", "path": "/Users/lpatiny/git/zakodium/molecules-parser/data/database/gpt-4-1106-preview-0.2-0.2_v200" }, "model": { "model": "gpt-4-1106-preview", "temperature": 0.2, "top_p": 0.2, "promptDescription": "Our baseline prompt" }, "stats": { "nmr": { "proton": { "nbSpectra": 597, "frequency": { "50": 4, "80": 2, "200": 2, "300": 11, "400": 18, "500": 4, "not specified": 5, "N/A": 1 }, "solvent": { "CDCl3": 95, "not specified": 6, "CD3OD": 35, "CHCl3": 1, "DMSO-d6": 73, "CD3COCD3": 2, "CD3CN": 6, "CDC13": 1, "pyridine-d5": 7, "MeOH": 4, "N/A": 1, "MeOD": 3, "chloroform": 1, "D2O": 2, "d4-MeOH": 37, "CDCl3 + d4-MeOH": 2, "see Table 1": 1, "DMSO": 1, "see Table 2": 1, "acetone-d6": 3 } }, "carbon": { "nbSpectra": 451, "frequency": { "50": 2, "75": 5, "80": 2, "100": 14, "125": 3, "not specified": 2, "50.3": 1, "75.4": 1 }, "solvent": { "DMSO-d6": 52, "not specified": 3, "CD3OD": 33, "CHCl3": 1, "CDCl3": 72, "CD3COCD3": 1, "pyridine-d5": 7, "MeOH": 4, "CD3CN": 4, "MeOD": 3, "DMSO": 2, "d4-MeOH": 35, "CDCl3 + d4-MeOH": 2, "see Table 1": 2, "acetone-d6": 5 } } }, "yaml": { "parsable": 637, "notParsable": 5, "empty": 297 }, "numbers": { "inventedNumbers": [ { "number": 100, "counter": 4 }, { "number": 31, "counter": 2 }, { "number": -18, "counter": 2 }, { "number": 54.08, "counter": 2 }, { "number": 23, "counter": 1 }, { "number": 24, "counter": 1 }, { "number": 29, "counter": 1 }, { "number": 30, "counter": 1 }, { "number": 102, "counter": 1 }, { "number": 126, "counter": 1 }, { "number": 130, "counter": 1 }, { "number": 1181, "counter": 1 }, { "number": 1275, "counter": 1 }, { "number": 1380, "counter": 1 }, { "number": 1505, "counter": 1 }, { "number": 1510, "counter": 1 }, { "number": 1518, "counter": 1 }, { "number": 1550, "counter": 1 }, { "number": 1604, "counter": 1 }, { "number": 1628, "counter": 1 }, { "number": 1630, "counter": 1 }, { "number": 1670, "counter": 1 }, { "number": 1689, "counter": 1 }, { "number": 7.69, "counter": 1 }, { "number": 10.61, "counter": 1 }, { "number": 8.08, "counter": 1 }, { "number": 129.55, "counter": 1 }, { "number": 33.3, "counter": 1 }, { "number": 33.53, "counter": 1 }, { "number": 7.14, "counter": 1 }, { "number": 7.16, "counter": 1 }, { "number": 1.18, "counter": 1 }, { "number": 2.71, "counter": 1 }, { "number": 296.5, "counter": 1 }, { "number": 3.77, "counter": 1 }, { "number": 7.91, "counter": 1 }, { "number": -23.86, "counter": 1 }, { "number": 3.8, "counter": 1 }, { "number": 3.83, "counter": 1 }, { "number": 54.06, "counter": 1 }, { "number": 7.52, "counter": 1 }, { "number": 7.5, "counter": 1 }, { "number": 8.5, "counter": 1 }, { "number": 4.45, "counter": 1 }, { "number": 149.8, "counter": 1 }, { "number": 115.4, "counter": 1 } ], "notParsedNumbers": [ { "number": 13, "counter": 417 }, { "number": -1, "counter": 175 }, { "number": 1, "counter": 53 }, { "number": 16, "counter": 23 }, { "number": 4, "counter": 22 }, { "number": 5, "counter": 22 }, { "number": 3, "counter": 20 }, { "number": 7, "counter": 19 }, { "number": 14, "counter": 19 }, { "number": 8, "counter": 17 }, { "number": 24, "counter": 17 }, { "number": 15, "counter": 16 }, { "number": 9, "counter": 15 }, { "number": 10, "counter": 15 }, { "number": 17, "counter": 15 }, { "number": 18, "counter": 14 }, { "number": 23, "counter": 14 }, { "number": 2, "counter": 13 }, { "number": 6, "counter": 12 }, { "number": 12, "counter": 12 }, { "number": 11, "counter": 11 }, { "number": 25, "counter": 11 }, { "number": 20, "counter": 10 }, { "number": 19, "counter": 9 }, { "number": 135, "counter": 9 }, { "number": 21, "counter": 8 }, { "number": 28, "counter": 8 }, { "number": 22, "counter": 7 }, { "number": 70, "counter": 7 }, { "number": 0, "counter": 6 }, { "number": 27, "counter": 6 }, { "number": 240, "counter": 6 }, { "number": -5, "counter": 6 }, { "number": 4.2, "counter": 6 }, { "number": 34, "counter": 5 }, { "number": 209, "counter": 5 }, { "number": 210, "counter": 5 }, { "number": 0.1, "counter": 5 }, { "number": 30, "counter": 4 }, { "number": 230, "counter": 4 }, { "number": -8, "counter": 4 }, { "number": -7, "counter": 4 }, { "number": -2, "counter": 4 }, { "number": 230.2, "counter": 4 }, { "number": 269.4, "counter": 4 }, { "number": 269.8, "counter": 4 }, { "number": 361.2, "counter": 4 }, { "number": 269.3, "counter": 4 }, { "number": 29, "counter": 3 }, { "number": 33, "counter": 3 } ], "valid": 588, "invalid": 54, "ratioPresent": 0.9385453337312842, "nbUnion": 35330, "duplicates": 44, "invented": 36 }, "meta": { "withName": 631, "withoutName": 11, "withMolfile": 476 }, "info": { "protonNMR": { "present": 604, "notPresent": 38, "valid": 272, "notValid": 332 }, "carbonNMR": { "present": 458, "notPresent": 184, "valid": 185, "notValid": 273 }, "elementalAnalysis": { "present": 261, "notPresent": 381, "valid": 223, "notValid": 38 }, "hrms": { "present": 63, "notPresent": 579, "valid": 14, "notValid": 49 } }, "logs": { "info": 97, "warn": 243, "error": 136, "fatal": 166 }, "chatgpt": { "promptTokens": 488935, "completionTokens": 449341, "totalTokens": 938276, "nbPrompts": 642, "version": "v200", "model": "gpt-4-1106-preview", "temperature": 0.2, "top_p": 0.2, "cost": 18.36958 } }, "description": "Our baseline prompt", "key": "Our baseline prompt_gpt-4-1106-preview_0.2_0.2", "version": "v200" } ], "versions": [ "v200" ], "stats": { "yaml": { "parsable": [ 637 ], "notParsable": [ 5 ], "empty": [ 297 ] }, "numbers": { "valid": [ 588 ], "invalid": [ 54 ], "ratioPresent": [ 0.9385453337312842 ], "nbUnion": [ 35330 ], "duplicates": [ 44 ], "invented": [ 36 ] }, "meta": { "withName": [ 631 ], "withoutName": [ 11 ], "withMolfile": [ 476 ] } }, "description": "Our baseline prompt" } ]