KapiCoder-Mongo-Nano / evaluation /val_v2c_summary.json
esilva's picture
Add v3 weights (v0.1.0-preview)
c03bfe1 verified
Raw History Blame Contribute Delete
20.6 kB
{
"schema": "kapi-hf-eval-summary/1",
"kind": "val_v2c",
"label": "Internal evaluation on synthetic data; no blind human-written set. Aggregates only; no item text.",
"model": "esilva/KapiCoder-Mongo-Nano",
"split": "val",
"n": 2000,
"seed": 20261008,
"contract": "editor-context-v1",
"dataset": {
"sample_ids_sha256": "28ba86be496839a321de0035d6a7e21827a5e8de4362151d80958ee8c798562f",
"sha256": "af3331481966f3ebfb05d36109b43014cd850f669f6f5d2a524f8da6e9609ec7",
"manifest_sha256": "2fef4f342d8a21eaf6d7e9eb1f223b3c8094d56c8dcd00e64f04bc20b20cd3cb",
"stratify_by": [
"cut_kind",
"domain"
],
"cut_kind": {
"multiline": 161,
"key": 319,
"value": 280,
"rest": 241,
"operator": 268,
"member": 201,
"stage": 251,
"partial": 200,
"stmt_start": 79
},
"domain": {
"commerce": 168,
"content": 161,
"energy": 171,
"finance": 175,
"gaming": 163,
"healthcare-synthetic": 173,
"hr": 168,
"iot": 168,
"logistics": 157,
"saas": 165,
"travel": 162,
"workspace": 169
}
},
"decoding": "greedy (argmax), padding do vocab mascarado",
"max_new_tokens": 96,
"context_tokens": 2048,
"load": "bf16",
"this_model": {
"headline": {
"n": 2000,
"exact_match": 0.405,
"normalized_match": 0.405,
"prefix_match": 0.4105,
"edit_similarity": 0.6499,
"chars_accepted": 7.9775,
"tab_chars": 6.343,
"accepted_prefix_tokens": 2.62,
"syntax_valid": 0.9955,
"mongo_valid": 0.992,
"exec_valid": 0.5945,
"exec_valid_incl_empty": 0.5945,
"nonexistent_fields_rate": 0.0265,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.997,
"mean_generated_tokens": 15.292
},
"headline_ide32": {
"n": 2000,
"exact_match": 0.4035,
"normalized_match": 0.4035,
"prefix_match": 0.4105,
"edit_similarity": 0.6421,
"chars_accepted": 7.9575,
"tab_chars": 6.323,
"accepted_prefix_tokens": 2.6125,
"syntax_valid": 0.836,
"mongo_valid": 0.833,
"exec_valid": 0.5055,
"exec_valid_incl_empty": 0.5055,
"nonexistent_fields_rate": 0.0165,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": null,
"mean_generated_tokens": null
},
"reference_ceiling": {
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6255,
"nonexistent_fields_rate": 0.0
},
"exec_stage_distribution": {
"ok": 1189,
"api": 483,
"syntax": 248,
"fields": 75,
"exec": 2,
"type": 3
},
"by_cut_kind": {
"key": {
"n": 319,
"exact_match": 0.4639,
"normalized_match": 0.4639,
"prefix_match": 0.4639,
"edit_similarity": 0.6428,
"chars_accepted": 6.2351,
"tab_chars": 5.7335,
"accepted_prefix_tokens": 1.9154,
"syntax_valid": 0.9969,
"mongo_valid": 0.9937,
"exec_valid": 0.5392,
"exec_valid_incl_empty": 0.5392,
"nonexistent_fields_rate": 0.0376,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 8.7147
},
"member": {
"n": 201,
"exact_match": 0.7711,
"normalized_match": 0.7711,
"prefix_match": 0.7761,
"edit_similarity": 0.8523,
"chars_accepted": 7.9254,
"tab_chars": 7.6368,
"accepted_prefix_tokens": 1.6269,
"syntax_valid": 1.0,
"mongo_valid": 0.995,
"exec_valid": 0.6368,
"exec_valid_incl_empty": 0.6368,
"nonexistent_fields_rate": 0.0,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 5.209
},
"multiline": {
"n": 161,
"exact_match": 0.0,
"normalized_match": 0.0,
"prefix_match": 0.0,
"edit_similarity": 0.4178,
"chars_accepted": 11.5155,
"tab_chars": 7.1491,
"accepted_prefix_tokens": 3.5839,
"syntax_valid": 0.9938,
"mongo_valid": 0.9938,
"exec_valid": 0.5901,
"exec_valid_incl_empty": 0.5901,
"nonexistent_fields_rate": 0.087,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9876,
"mean_generated_tokens": 43.7578
},
"operator": {
"n": 268,
"exact_match": 0.6045,
"normalized_match": 0.6045,
"prefix_match": 0.6157,
"edit_similarity": 0.7894,
"chars_accepted": 7.6194,
"tab_chars": 6.4701,
"accepted_prefix_tokens": 3.0261,
"syntax_valid": 0.9963,
"mongo_valid": 0.9963,
"exec_valid": 0.6306,
"exec_valid_incl_empty": 0.6306,
"nonexistent_fields_rate": 0.0187,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9963,
"mean_generated_tokens": 9.8881
},
"partial": {
"n": 200,
"exact_match": 0.695,
"normalized_match": 0.695,
"prefix_match": 0.695,
"edit_similarity": 0.8264,
"chars_accepted": 6.34,
"tab_chars": 5.425,
"accepted_prefix_tokens": 2.405,
"syntax_valid": 0.995,
"mongo_valid": 0.98,
"exec_valid": 0.57,
"exec_valid_incl_empty": 0.57,
"nonexistent_fields_rate": 0.05,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 7.9
},
"rest": {
"n": 241,
"exact_match": 0.3444,
"normalized_match": 0.3444,
"prefix_match": 0.3485,
"edit_similarity": 0.6293,
"chars_accepted": 9.4855,
"tab_chars": 7.3361,
"accepted_prefix_tokens": 2.9793,
"syntax_valid": 0.9959,
"mongo_valid": 0.9876,
"exec_valid": 0.5809,
"exec_valid_incl_empty": 0.5809,
"nonexistent_fields_rate": 0.0124,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 16.7054
},
"stage": {
"n": 251,
"exact_match": 0.012,
"normalized_match": 0.012,
"prefix_match": 0.012,
"edit_similarity": 0.4255,
"chars_accepted": 11.4861,
"tab_chars": 7.8247,
"accepted_prefix_tokens": 4.243,
"syntax_valid": 0.992,
"mongo_valid": 0.992,
"exec_valid": 0.5817,
"exec_valid_incl_empty": 0.5817,
"nonexistent_fields_rate": 0.0239,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.992,
"mean_generated_tokens": 30.7131
},
"stmt_start": {
"n": 79,
"exact_match": 0.3671,
"normalized_match": 0.3671,
"prefix_match": 0.3671,
"edit_similarity": 0.6448,
"chars_accepted": 16.0633,
"tab_chars": 13.8734,
"accepted_prefix_tokens": 3.5316,
"syntax_valid": 0.9873,
"mongo_valid": 0.9873,
"exec_valid": 0.7468,
"exec_valid_incl_empty": 0.7468,
"nonexistent_fields_rate": 0.038,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9873,
"mean_generated_tokens": 16.3924
},
"value": {
"n": 280,
"exact_match": 0.325,
"normalized_match": 0.325,
"prefix_match": 0.3464,
"edit_similarity": 0.6071,
"chars_accepted": 2.7536,
"tab_chars": 1.8714,
"accepted_prefix_tokens": 1.325,
"syntax_valid": 0.9964,
"mongo_valid": 0.9964,
"exec_valid": 0.5929,
"exec_valid_incl_empty": 0.5929,
"nonexistent_fields_rate": 0.0,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 8.7571
}
},
"by_domain": {
"commerce": {
"n": 168,
"exact_match": 0.375,
"normalized_match": 0.375,
"prefix_match": 0.381,
"edit_similarity": 0.623,
"chars_accepted": 6.9881,
"tab_chars": 5.3214,
"accepted_prefix_tokens": 2.25,
"syntax_valid": 1.0,
"mongo_valid": 0.994,
"exec_valid": 0.6548,
"exec_valid_incl_empty": 0.6548,
"nonexistent_fields_rate": 0.0298,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 14.1786
},
"content": {
"n": 161,
"exact_match": 0.3913,
"normalized_match": 0.3913,
"prefix_match": 0.4099,
"edit_similarity": 0.6281,
"chars_accepted": 8.0497,
"tab_chars": 6.5652,
"accepted_prefix_tokens": 2.7019,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.5901,
"exec_valid_incl_empty": 0.5901,
"nonexistent_fields_rate": 0.0062,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 14.8012
},
"energy": {
"n": 171,
"exact_match": 0.4269,
"normalized_match": 0.4269,
"prefix_match": 0.4327,
"edit_similarity": 0.6494,
"chars_accepted": 7.7953,
"tab_chars": 6.5146,
"accepted_prefix_tokens": 2.3509,
"syntax_valid": 0.9942,
"mongo_valid": 0.9942,
"exec_valid": 0.614,
"exec_valid_incl_empty": 0.614,
"nonexistent_fields_rate": 0.0351,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9942,
"mean_generated_tokens": 16.6842
},
"finance": {
"n": 175,
"exact_match": 0.3486,
"normalized_match": 0.3486,
"prefix_match": 0.3543,
"edit_similarity": 0.6135,
"chars_accepted": 8.1257,
"tab_chars": 6.6286,
"accepted_prefix_tokens": 2.88,
"syntax_valid": 0.9943,
"mongo_valid": 0.9943,
"exec_valid": 0.5943,
"exec_valid_incl_empty": 0.5943,
"nonexistent_fields_rate": 0.04,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 18.5257
},
"gaming": {
"n": 163,
"exact_match": 0.3681,
"normalized_match": 0.3681,
"prefix_match": 0.3804,
"edit_similarity": 0.6296,
"chars_accepted": 7.6196,
"tab_chars": 6.0184,
"accepted_prefix_tokens": 2.5092,
"syntax_valid": 0.9877,
"mongo_valid": 0.9755,
"exec_valid": 0.5951,
"exec_valid_incl_empty": 0.5951,
"nonexistent_fields_rate": 0.0245,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 14.8712
},
"healthcare-synthetic": {
"n": 173,
"exact_match": 0.4104,
"normalized_match": 0.4104,
"prefix_match": 0.4104,
"edit_similarity": 0.6618,
"chars_accepted": 8.3815,
"tab_chars": 6.4624,
"accepted_prefix_tokens": 2.7341,
"syntax_valid": 0.9942,
"mongo_valid": 0.9884,
"exec_valid": 0.5607,
"exec_valid_incl_empty": 0.5607,
"nonexistent_fields_rate": 0.0231,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9942,
"mean_generated_tokens": 15.0636
},
"hr": {
"n": 168,
"exact_match": 0.4643,
"normalized_match": 0.4643,
"prefix_match": 0.4762,
"edit_similarity": 0.6899,
"chars_accepted": 7.8155,
"tab_chars": 6.3393,
"accepted_prefix_tokens": 2.5357,
"syntax_valid": 0.994,
"mongo_valid": 0.994,
"exec_valid": 0.5833,
"exec_valid_incl_empty": 0.5833,
"nonexistent_fields_rate": 0.0119,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.994,
"mean_generated_tokens": 12.7738
},
"iot": {
"n": 168,
"exact_match": 0.4107,
"normalized_match": 0.4107,
"prefix_match": 0.4107,
"edit_similarity": 0.6606,
"chars_accepted": 8.4821,
"tab_chars": 6.4643,
"accepted_prefix_tokens": 2.869,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6369,
"exec_valid_incl_empty": 0.6369,
"nonexistent_fields_rate": 0.0536,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 15.3155
},
"logistics": {
"n": 157,
"exact_match": 0.4076,
"normalized_match": 0.4076,
"prefix_match": 0.4076,
"edit_similarity": 0.6721,
"chars_accepted": 8.9554,
"tab_chars": 6.8408,
"accepted_prefix_tokens": 2.8535,
"syntax_valid": 1.0,
"mongo_valid": 0.9936,
"exec_valid": 0.5987,
"exec_valid_incl_empty": 0.5987,
"nonexistent_fields_rate": 0.0255,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 15.2102
},
"saas": {
"n": 165,
"exact_match": 0.4788,
"normalized_match": 0.4788,
"prefix_match": 0.4788,
"edit_similarity": 0.7021,
"chars_accepted": 7.0242,
"tab_chars": 5.5636,
"accepted_prefix_tokens": 2.3758,
"syntax_valid": 0.9939,
"mongo_valid": 0.9939,
"exec_valid": 0.4848,
"exec_valid_incl_empty": 0.4848,
"nonexistent_fields_rate": 0.0364,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 12.4909
},
"travel": {
"n": 162,
"exact_match": 0.4136,
"normalized_match": 0.4136,
"prefix_match": 0.4198,
"edit_similarity": 0.6412,
"chars_accepted": 8.4506,
"tab_chars": 6.6667,
"accepted_prefix_tokens": 2.7901,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6543,
"exec_valid_incl_empty": 0.6543,
"nonexistent_fields_rate": 0.0123,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 16.5123
},
"workspace": {
"n": 169,
"exact_match": 0.3669,
"normalized_match": 0.3669,
"prefix_match": 0.3669,
"edit_similarity": 0.6296,
"chars_accepted": 8.0828,
"tab_chars": 6.7396,
"accepted_prefix_tokens": 2.5976,
"syntax_valid": 0.9882,
"mongo_valid": 0.9763,
"exec_valid": 0.568,
"exec_valid_incl_empty": 0.568,
"nonexistent_fields_rate": 0.0178,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9822,
"mean_generated_tokens": 16.8698
}
},
"by_context_kind": {
"full": {
"n": 1647,
"exact_match": 0.4141,
"normalized_match": 0.4141,
"prefix_match": 0.4196,
"edit_similarity": 0.6598,
"chars_accepted": 8.1488,
"tab_chars": 6.4584,
"accepted_prefix_tokens": 2.6752,
"syntax_valid": 0.9951,
"mongo_valid": 0.9927,
"exec_valid": 0.6169,
"exec_valid_incl_empty": 0.6169,
"nonexistent_fields_rate": 0.0134,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.997,
"mean_generated_tokens": 15.4244
},
"irrelevant": {
"n": 198,
"exact_match": 0.3687,
"normalized_match": 0.3687,
"prefix_match": 0.3737,
"edit_similarity": 0.5935,
"chars_accepted": 7.0707,
"tab_chars": 5.8081,
"accepted_prefix_tokens": 2.2374,
"syntax_valid": 1.0,
"mongo_valid": 0.9949,
"exec_valid": 0.5051,
"exec_valid_incl_empty": 0.5051,
"nonexistent_fields_rate": 0.0859,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9949,
"mean_generated_tokens": 14.5505
},
"none": {
"n": 155,
"exact_match": 0.3548,
"normalized_match": 0.3548,
"prefix_match": 0.3613,
"edit_similarity": 0.6167,
"chars_accepted": 7.3161,
"tab_chars": 5.8,
"accepted_prefix_tokens": 2.5226,
"syntax_valid": 0.9935,
"mongo_valid": 0.9806,
"exec_valid": 0.471,
"exec_valid_incl_empty": 0.471,
"nonexistent_fields_rate": 0.0903,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 14.8323
}
}
},
"metrics_by_model": {
"this_model_v3": {
"syntax_valid": 0.9955,
"mongodb_valid": 0.992,
"nonexistent_fields_rate": 2.65,
"conversational_suggestion_rate": 0.0,
"fim_acc": 0.405,
"exact_match": 0.405,
"apt_mean": 2.62,
"exec_valid_autocomplete": 0.5945,
"empty_rate": 0.0,
"reference_ceiling_exec_valid": 0.6255,
"exec_valid_vs_ceiling": 0.9504,
"nonexistent_fields_full_rate_pct": 1.34,
"n_context_full": 1647,
"exec_valid_full": 0.6169,
"nonexistent_fields_irrelevant_rate_pct": 8.59,
"n_context_irrelevant": 198,
"exec_valid_irrelevant": 0.5051,
"nonexistent_fields_none_rate_pct": 9.03,
"n_context_none": 155,
"exec_valid_none": 0.471,
"info_fim_acc_gain_vs_zero_shot_base_pp": 32.9,
"info_apt_delta_vs_zero_shot_base": 1.4155,
"info_exec_valid_delta_vs_zero_shot_base_pp": 38.15,
"generic_js_fim_acc": 0.53,
"generic_js_fim_delta_pp_vs_zero_shot_base": 19.0,
"write_or_destructive_executions": 0,
"train_eval_leak_count": 0,
"decontamination_approved": true,
"auto_report_attached": true
},
"previous_internal_checkpoint_v1_not_released": {
"syntax_valid": 0.995,
"mongodb_valid": 0.994,
"nonexistent_fields_rate": 3.8,
"conversational_suggestion_rate": 0.0,
"fim_acc": 0.403,
"exact_match": 0.403,
"apt_mean": 2.6605,
"exec_valid_autocomplete": 0.582,
"empty_rate": 0.0,
"reference_ceiling_exec_valid": 0.6255,
"exec_valid_vs_ceiling": 0.9305,
"nonexistent_fields_full_rate_pct": 2.06,
"n_context_full": 1647,
"exec_valid_full": 0.6084,
"nonexistent_fields_irrelevant_rate_pct": 11.11,
"n_context_irrelevant": 198,
"exec_valid_irrelevant": 0.4798,
"nonexistent_fields_none_rate_pct": 12.9,
"n_context_none": 155,
"exec_valid_none": 0.4323,
"info_fim_acc_gain_vs_zero_shot_base_pp": 32.7,
"info_apt_delta_vs_zero_shot_base": 1.456,
"info_exec_valid_delta_vs_zero_shot_base_pp": 36.9,
"generic_js_fim_acc": 0.518,
"generic_js_fim_delta_pp_vs_zero_shot_base": 17.8,
"write_or_destructive_executions": 0,
"train_eval_leak_count": 0,
"decontamination_approved": true,
"auto_report_attached": true
},
"zero_shot_base": {
"syntax_valid": 0.498,
"mongodb_valid": 0.4275,
"nonexistent_fields_rate": 4.85,
"conversational_suggestion_rate": 0.0,
"fim_acc": 0.076,
"exact_match": 0.076,
"apt_mean": 1.2045,
"exec_valid_autocomplete": 0.213,
"empty_rate": 0.0055,
"reference_ceiling_exec_valid": 0.6255,
"exec_valid_vs_ceiling": 0.3405,
"nonexistent_fields_full_rate_pct": 4.13,
"n_context_full": 1647,
"exec_valid_full": 0.2198,
"nonexistent_fields_irrelevant_rate_pct": 8.59,
"n_context_irrelevant": 198,
"exec_valid_irrelevant": 0.1667,
"nonexistent_fields_none_rate_pct": 7.74,
"n_context_none": 155,
"exec_valid_none": 0.2,
"generic_js_fim_acc": 0.34,
"write_or_destructive_executions": 0,
"train_eval_leak_count": 0,
"decontamination_approved": true,
"auto_report_attached": true
}
},
"metric_definitions": {
"exact_match": "pred.strip() == middle.strip()",
"normalized_match": "igualdade com espacos normalizados (FIM accuracy do bake-off)",
"chars_accepted": "tamanho do prefixo comum (caracteres) entre predicao e middle",
"accepted_prefix_tokens": "tokens iniciais da referencia cobertos pelo prefixo comum (APT)",
"syntax_valid": "node vm.Script sobre prefix+pred+suffix",
"mongo_valid": "mongo_validate (AST, estatico) sobre o documento reconstruido",
"exec_valid": "chat_validate.validate_text stage=ok (sintaxe, API do Console, campos, tipos, execucao somente-leitura em kapi_fx_*) E predicao nao vazia; exec_valid_incl_empty conta vazio como valido",
"nonexistent_fields_rate": "stage=fields com 'campo/estrutura invalido' (campo ausente do schema) sobre todos os exemplos",
"reference_ceiling": "mesmas checagens aplicadas ao middle de referencia (teto/calibracao do validador)"
}
}