KapiCoder-Mongo-Small / evaluation /val_v2c_summary.json
esilva's picture
Add v3 weights (v0.1.0-preview)
4ba57bc 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-Small",
"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": 0.41,
"normalized_match": 0.41,
"prefix_match": 0.416,
"edit_similarity": 0.6536,
"chars_accepted": 8.214,
"tab_chars": 6.5425,
"accepted_prefix_tokens": 2.698,
"syntax_valid": 0.998,
"mongo_valid": 0.9975,
"exec_valid": 0.6015,
"exec_valid_incl_empty": 0.6015,
"nonexistent_fields_rate": 0.021,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9975,
"mean_generated_tokens": 15.3875
},
"headline_ide32": {
"n": 2000,
"exact_match": 0.409,
"normalized_match": 0.409,
"prefix_match": 0.4165,
"edit_similarity": 0.6463,
"chars_accepted": 8.2,
"tab_chars": 6.5285,
"accepted_prefix_tokens": 2.693,
"syntax_valid": 0.836,
"mongo_valid": 0.8355,
"exec_valid": 0.5115,
"exec_valid_incl_empty": 0.5115,
"nonexistent_fields_rate": 0.012,
"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": 1203,
"api": 490,
"syntax": 243,
"type": 3,
"fields": 59,
"exec": 2
},
"by_cut_kind": {
"key": {
"n": 319,
"exact_match": 0.4514,
"normalized_match": 0.4514,
"prefix_match": 0.4545,
"edit_similarity": 0.6407,
"chars_accepted": 6.4639,
"tab_chars": 5.8119,
"accepted_prefix_tokens": 2.0,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.5361,
"exec_valid_incl_empty": 0.5361,
"nonexistent_fields_rate": 0.0408,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 9.1442
},
"member": {
"n": 201,
"exact_match": 0.801,
"normalized_match": 0.801,
"prefix_match": 0.801,
"edit_similarity": 0.8754,
"chars_accepted": 8.6965,
"tab_chars": 8.3134,
"accepted_prefix_tokens": 1.8259,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6517,
"exec_valid_incl_empty": 0.6517,
"nonexistent_fields_rate": 0.0,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 5.1045
},
"multiline": {
"n": 161,
"exact_match": 0.0,
"normalized_match": 0.0,
"prefix_match": 0.0,
"edit_similarity": 0.4159,
"chars_accepted": 11.5652,
"tab_chars": 7.2671,
"accepted_prefix_tokens": 3.677,
"syntax_valid": 0.9876,
"mongo_valid": 0.9876,
"exec_valid": 0.6149,
"exec_valid_incl_empty": 0.6149,
"nonexistent_fields_rate": 0.0807,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9814,
"mean_generated_tokens": 44.3665
},
"operator": {
"n": 268,
"exact_match": 0.6119,
"normalized_match": 0.6119,
"prefix_match": 0.6231,
"edit_similarity": 0.7941,
"chars_accepted": 7.7052,
"tab_chars": 6.5522,
"accepted_prefix_tokens": 3.0597,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6306,
"exec_valid_incl_empty": 0.6306,
"nonexistent_fields_rate": 0.0037,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 9.8134
},
"partial": {
"n": 200,
"exact_match": 0.72,
"normalized_match": 0.72,
"prefix_match": 0.72,
"edit_similarity": 0.8414,
"chars_accepted": 6.265,
"tab_chars": 5.435,
"accepted_prefix_tokens": 2.36,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.59,
"exec_valid_incl_empty": 0.59,
"nonexistent_fields_rate": 0.035,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 7.66
},
"rest": {
"n": 241,
"exact_match": 0.3444,
"normalized_match": 0.3444,
"prefix_match": 0.3485,
"edit_similarity": 0.6199,
"chars_accepted": 9.8714,
"tab_chars": 7.7344,
"accepted_prefix_tokens": 3.1286,
"syntax_valid": 1.0,
"mongo_valid": 0.9959,
"exec_valid": 0.5892,
"exec_valid_incl_empty": 0.5892,
"nonexistent_fields_rate": 0.0124,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 17.5021
},
"stage": {
"n": 251,
"exact_match": 0.0239,
"normalized_match": 0.0239,
"prefix_match": 0.0239,
"edit_similarity": 0.4418,
"chars_accepted": 11.8367,
"tab_chars": 8.1474,
"accepted_prefix_tokens": 4.3347,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.5976,
"exec_valid_incl_empty": 0.5976,
"nonexistent_fields_rate": 0.012,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 30.0478
},
"stmt_start": {
"n": 79,
"exact_match": 0.3671,
"normalized_match": 0.3671,
"prefix_match": 0.3671,
"edit_similarity": 0.6481,
"chars_accepted": 16.2785,
"tab_chars": 13.9747,
"accepted_prefix_tokens": 3.6076,
"syntax_valid": 0.9873,
"mongo_valid": 0.9873,
"exec_valid": 0.7089,
"exec_valid_incl_empty": 0.7089,
"nonexistent_fields_rate": 0.0253,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9873,
"mean_generated_tokens": 16.3924
},
"value": {
"n": 280,
"exact_match": 0.3179,
"normalized_match": 0.3179,
"prefix_match": 0.3429,
"edit_similarity": 0.5979,
"chars_accepted": 2.8643,
"tab_chars": 1.9071,
"accepted_prefix_tokens": 1.3571,
"syntax_valid": 0.9964,
"mongo_valid": 0.9964,
"exec_valid": 0.5964,
"exec_valid_incl_empty": 0.5964,
"nonexistent_fields_rate": 0.0,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9964,
"mean_generated_tokens": 8.8286
}
},
"by_domain": {
"commerce": {
"n": 168,
"exact_match": 0.381,
"normalized_match": 0.381,
"prefix_match": 0.3869,
"edit_similarity": 0.6404,
"chars_accepted": 7.1667,
"tab_chars": 5.4464,
"accepted_prefix_tokens": 2.2917,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6607,
"exec_valid_incl_empty": 0.6607,
"nonexistent_fields_rate": 0.0119,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 15.1488
},
"content": {
"n": 161,
"exact_match": 0.3851,
"normalized_match": 0.3851,
"prefix_match": 0.4037,
"edit_similarity": 0.6223,
"chars_accepted": 8.1366,
"tab_chars": 6.5093,
"accepted_prefix_tokens": 2.6894,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6087,
"exec_valid_incl_empty": 0.6087,
"nonexistent_fields_rate": 0.0062,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 15.1491
},
"energy": {
"n": 171,
"exact_match": 0.4386,
"normalized_match": 0.4386,
"prefix_match": 0.4444,
"edit_similarity": 0.6549,
"chars_accepted": 8.1637,
"tab_chars": 6.8129,
"accepted_prefix_tokens": 2.4737,
"syntax_valid": 0.9942,
"mongo_valid": 0.9942,
"exec_valid": 0.6023,
"exec_valid_incl_empty": 0.6023,
"nonexistent_fields_rate": 0.0468,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9942,
"mean_generated_tokens": 16.0526
},
"finance": {
"n": 175,
"exact_match": 0.3429,
"normalized_match": 0.3429,
"prefix_match": 0.3486,
"edit_similarity": 0.6153,
"chars_accepted": 8.2457,
"tab_chars": 6.5314,
"accepted_prefix_tokens": 2.9486,
"syntax_valid": 0.9943,
"mongo_valid": 0.9943,
"exec_valid": 0.5943,
"exec_valid_incl_empty": 0.5943,
"nonexistent_fields_rate": 0.0286,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9943,
"mean_generated_tokens": 18.6514
},
"gaming": {
"n": 163,
"exact_match": 0.3742,
"normalized_match": 0.3742,
"prefix_match": 0.3804,
"edit_similarity": 0.6392,
"chars_accepted": 7.9264,
"tab_chars": 6.1902,
"accepted_prefix_tokens": 2.5644,
"syntax_valid": 1.0,
"mongo_valid": 0.9939,
"exec_valid": 0.6012,
"exec_valid_incl_empty": 0.6012,
"nonexistent_fields_rate": 0.0184,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 15.0184
},
"healthcare-synthetic": {
"n": 173,
"exact_match": 0.4046,
"normalized_match": 0.4046,
"prefix_match": 0.4046,
"edit_similarity": 0.6652,
"chars_accepted": 8.5896,
"tab_chars": 6.4913,
"accepted_prefix_tokens": 2.8497,
"syntax_valid": 0.9942,
"mongo_valid": 0.9942,
"exec_valid": 0.5838,
"exec_valid_incl_empty": 0.5838,
"nonexistent_fields_rate": 0.0116,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9942,
"mean_generated_tokens": 14.9827
},
"hr": {
"n": 168,
"exact_match": 0.4762,
"normalized_match": 0.4762,
"prefix_match": 0.4881,
"edit_similarity": 0.7032,
"chars_accepted": 8.8333,
"tab_chars": 7.5119,
"accepted_prefix_tokens": 2.8512,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6012,
"exec_valid_incl_empty": 0.6012,
"nonexistent_fields_rate": 0.0119,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 12.8869
},
"iot": {
"n": 168,
"exact_match": 0.4345,
"normalized_match": 0.4345,
"prefix_match": 0.4405,
"edit_similarity": 0.6659,
"chars_accepted": 8.8393,
"tab_chars": 6.9583,
"accepted_prefix_tokens": 2.9702,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6429,
"exec_valid_incl_empty": 0.6429,
"nonexistent_fields_rate": 0.0357,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 15.6071
},
"logistics": {
"n": 157,
"exact_match": 0.4331,
"normalized_match": 0.4331,
"prefix_match": 0.4331,
"edit_similarity": 0.6879,
"chars_accepted": 8.9172,
"tab_chars": 6.8726,
"accepted_prefix_tokens": 2.8917,
"syntax_valid": 0.9936,
"mongo_valid": 0.9936,
"exec_valid": 0.586,
"exec_valid_incl_empty": 0.586,
"nonexistent_fields_rate": 0.0318,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9936,
"mean_generated_tokens": 14.5669
},
"saas": {
"n": 165,
"exact_match": 0.4788,
"normalized_match": 0.4788,
"prefix_match": 0.4848,
"edit_similarity": 0.6938,
"chars_accepted": 7.2424,
"tab_chars": 5.6788,
"accepted_prefix_tokens": 2.4909,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.5091,
"exec_valid_incl_empty": 0.5091,
"nonexistent_fields_rate": 0.0303,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 12.6061
},
"travel": {
"n": 162,
"exact_match": 0.4074,
"normalized_match": 0.4074,
"prefix_match": 0.4136,
"edit_similarity": 0.6385,
"chars_accepted": 8.8086,
"tab_chars": 7.0864,
"accepted_prefix_tokens": 2.9136,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.6605,
"exec_valid_incl_empty": 0.6605,
"nonexistent_fields_rate": 0.0062,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 16.5309
},
"workspace": {
"n": 169,
"exact_match": 0.3669,
"normalized_match": 0.3669,
"prefix_match": 0.3669,
"edit_similarity": 0.6191,
"chars_accepted": 7.7278,
"tab_chars": 6.432,
"accepted_prefix_tokens": 2.4438,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.568,
"exec_valid_incl_empty": 0.568,
"nonexistent_fields_rate": 0.0118,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9941,
"mean_generated_tokens": 17.2189
}
},
"by_context_kind": {
"full": {
"n": 1647,
"exact_match": 0.4183,
"normalized_match": 0.4183,
"prefix_match": 0.4238,
"edit_similarity": 0.6613,
"chars_accepted": 8.3734,
"tab_chars": 6.6758,
"accepted_prefix_tokens": 2.759,
"syntax_valid": 0.9976,
"mongo_valid": 0.997,
"exec_valid": 0.6211,
"exec_valid_incl_empty": 0.6211,
"nonexistent_fields_rate": 0.0121,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9976,
"mean_generated_tokens": 15.4857
},
"irrelevant": {
"n": 198,
"exact_match": 0.3838,
"normalized_match": 0.3838,
"prefix_match": 0.3889,
"edit_similarity": 0.6152,
"chars_accepted": 7.5657,
"tab_chars": 6.2121,
"accepted_prefix_tokens": 2.3333,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.5303,
"exec_valid_incl_empty": 0.5303,
"nonexistent_fields_rate": 0.0556,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 0.9949,
"mean_generated_tokens": 15.0
},
"none": {
"n": 155,
"exact_match": 0.3548,
"normalized_match": 0.3548,
"prefix_match": 0.3677,
"edit_similarity": 0.6217,
"chars_accepted": 7.3484,
"tab_chars": 5.5484,
"accepted_prefix_tokens": 2.5161,
"syntax_valid": 1.0,
"mongo_valid": 1.0,
"exec_valid": 0.4839,
"exec_valid_incl_empty": 0.4839,
"nonexistent_fields_rate": 0.071,
"empty_rate": 0.0,
"eos_or_suffix_stop_rate": 1.0,
"mean_generated_tokens": 14.8387
}
}
},
"metrics_by_model": {
"this_model_v3": {
"syntax_valid": 0.998,
"mongodb_valid": 0.9975,
"nonexistent_fields_rate": 2.1,
"conversational_suggestion_rate": 0.0,
"fim_acc": 0.41,
"exact_match": 0.41,
"apt_mean": 2.698,
"exec_valid_autocomplete": 0.6015,
"empty_rate": 0.0,
"reference_ceiling_exec_valid": 0.6255,
"exec_valid_vs_ceiling": 0.9616,
"nonexistent_fields_full_rate_pct": 1.21,
"n_context_full": 1647,
"exec_valid_full": 0.6211,
"nonexistent_fields_irrelevant_rate_pct": 5.56,
"n_context_irrelevant": 198,
"exec_valid_irrelevant": 0.5303,
"nonexistent_fields_none_rate_pct": 7.1,
"n_context_none": 155,
"exec_valid_none": 0.4839,
"info_fim_acc_gain_vs_zero_shot_base_pp": 26.8,
"info_apt_delta_vs_zero_shot_base": 1.259,
"info_exec_valid_delta_vs_zero_shot_base_pp": 28.55,
"generic_js_fim_acc": 0.606,
"generic_js_fim_delta_pp_vs_zero_shot_base": 20.0,
"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": 0.9955,
"mongodb_valid": 0.9945,
"nonexistent_fields_rate": 2.9,
"conversational_suggestion_rate": 0.0,
"fim_acc": 0.4105,
"exact_match": 0.4105,
"apt_mean": 2.6665,
"exec_valid_autocomplete": 0.595,
"empty_rate": 0.0,
"reference_ceiling_exec_valid": 0.6255,
"exec_valid_vs_ceiling": 0.9512,
"nonexistent_fields_full_rate_pct": 1.58,
"n_context_full": 1647,
"exec_valid_full": 0.6199,
"nonexistent_fields_irrelevant_rate_pct": 7.58,
"n_context_irrelevant": 198,
"exec_valid_irrelevant": 0.5051,
"nonexistent_fields_none_rate_pct": 10.97,
"n_context_none": 155,
"exec_valid_none": 0.4452,
"info_fim_acc_gain_vs_zero_shot_base_pp": 26.85,
"info_apt_delta_vs_zero_shot_base": 1.2275,
"info_exec_valid_delta_vs_zero_shot_base_pp": 27.9,
"generic_js_fim_acc": 0.632,
"generic_js_fim_delta_pp_vs_zero_shot_base": 22.6,
"write_or_destructive_executions": 0,
"train_eval_leak_count": 0,
"decontamination_approved": true,
"auto_report_attached": true
},
"zero_shot_base": {
"syntax_valid": 0.667,
"mongodb_valid": 0.595,
"nonexistent_fields_rate": 5.1,
"conversational_suggestion_rate": 0.0,
"fim_acc": 0.142,
"exact_match": 0.142,
"apt_mean": 1.439,
"exec_valid_autocomplete": 0.316,
"empty_rate": 0.0095,
"reference_ceiling_exec_valid": 0.6255,
"exec_valid_vs_ceiling": 0.5052,
"nonexistent_fields_full_rate_pct": 3.76,
"n_context_full": 1647,
"exec_valid_full": 0.3364,
"nonexistent_fields_irrelevant_rate_pct": 10.61,
"n_context_irrelevant": 198,
"exec_valid_irrelevant": 0.2172,
"nonexistent_fields_none_rate_pct": 12.26,
"n_context_none": 155,
"exec_valid_none": 0.2258,
"generic_js_fim_acc": 0.406,
"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)"
}
}