CognitiveMemoryLayer-models / extractor_training_metadata.json
avinashm's picture
Upload all trained models
e65379c verified
{
"family": "extractor",
"trained_at_utc": "2026-03-24T13:45:54.396697+00:00",
"rows": {
"train": 288000,
"test": 36000,
"eval": 36000
},
"tasks": [
"constraint_scope",
"constraint_stability",
"constraint_type",
"fact_type",
"pii_presence"
],
"task_label_counts": {
"train": {
"constraint_scope": {
"finance": 8000,
"food": 8000,
"general": 8000,
"health": 8000,
"none": 8000,
"social": 8000,
"tech": 8000,
"travel": 8000,
"work": 8000
},
"constraint_stability": {
"semi_stable": 8000,
"stable": 8000,
"volatile": 8000
},
"constraint_type": {
"causal": 8000,
"constraint_other": 8000,
"goal": 8000,
"none": 8000,
"policy": 8000,
"preference": 8000,
"state": 8000,
"value": 8000
},
"fact_type": {
"identity": 8000,
"location": 8000,
"none": 8000,
"occupation": 8000,
"other_fact": 8000,
"preference": 8000
},
"pii_presence": {
"no_pii": 40000,
"pii": 40000
}
},
"test": {
"constraint_scope": {
"finance": 1000,
"food": 1000,
"general": 1000,
"health": 1000,
"none": 1000,
"social": 1000,
"tech": 1000,
"travel": 1000,
"work": 1000
},
"constraint_stability": {
"semi_stable": 1000,
"stable": 1000,
"volatile": 1000
},
"constraint_type": {
"causal": 1000,
"constraint_other": 1000,
"goal": 1000,
"none": 1000,
"policy": 1000,
"preference": 1000,
"state": 1000,
"value": 1000
},
"fact_type": {
"identity": 1000,
"location": 1000,
"none": 1000,
"occupation": 1000,
"other_fact": 1000,
"preference": 1000
},
"pii_presence": {
"no_pii": 5000,
"pii": 5000
}
},
"eval": {
"constraint_scope": {
"finance": 1000,
"food": 1000,
"general": 1000,
"health": 1000,
"none": 1000,
"social": 1000,
"tech": 1000,
"travel": 1000,
"work": 1000
},
"constraint_stability": {
"semi_stable": 1000,
"stable": 1000,
"volatile": 1000
},
"constraint_type": {
"causal": 1000,
"constraint_other": 1000,
"goal": 1000,
"none": 1000,
"policy": 1000,
"preference": 1000,
"state": 1000,
"value": 1000
},
"fact_type": {
"identity": 1000,
"location": 1000,
"none": 1000,
"occupation": 1000,
"other_fact": 1000,
"preference": 1000
},
"pii_presence": {
"no_pii": 5000,
"pii": 5000
}
}
},
"labels": [
"constraint_scope::finance",
"constraint_scope::food",
"constraint_scope::general",
"constraint_scope::health",
"constraint_scope::none",
"constraint_scope::social",
"constraint_scope::tech",
"constraint_scope::travel",
"constraint_scope::work",
"constraint_stability::semi_stable",
"constraint_stability::stable",
"constraint_stability::volatile",
"constraint_type::causal",
"constraint_type::constraint_other",
"constraint_type::goal",
"constraint_type::none",
"constraint_type::policy",
"constraint_type::preference",
"constraint_type::state",
"constraint_type::value",
"fact_type::identity",
"fact_type::location",
"fact_type::none",
"fact_type::occupation",
"fact_type::other_fact",
"fact_type::preference",
"pii_presence::no_pii",
"pii_presence::pii"
],
"label_to_id": {
"constraint_scope::finance": 0,
"constraint_scope::food": 1,
"constraint_scope::general": 2,
"constraint_scope::health": 3,
"constraint_scope::none": 4,
"constraint_scope::social": 5,
"constraint_scope::tech": 6,
"constraint_scope::travel": 7,
"constraint_scope::work": 8,
"constraint_stability::semi_stable": 9,
"constraint_stability::stable": 10,
"constraint_stability::volatile": 11,
"constraint_type::causal": 12,
"constraint_type::constraint_other": 13,
"constraint_type::goal": 14,
"constraint_type::none": 15,
"constraint_type::policy": 16,
"constraint_type::preference": 17,
"constraint_type::state": 18,
"constraint_type::value": 19,
"fact_type::identity": 20,
"fact_type::location": 21,
"fact_type::none": 22,
"fact_type::occupation": 23,
"fact_type::other_fact": 24,
"fact_type::preference": 25,
"pii_presence::no_pii": 26,
"pii_presence::pii": 27
},
"train_config": {
"max_features": 250000,
"min_df": 2,
"ngram_min": 1,
"ngram_max": 2,
"max_iter": 25,
"alpha": 1e-05,
"seed": 42,
"predict_batch_size": 8192,
"early_stopping": true,
"early_stopping_patience": 3,
"early_stopping_metric": "macro_f1",
"early_stopping_min_delta": 0.001,
"calibration_method": "sigmoid"
},
"training_summary": {
"actual_epochs": 4,
"best_epoch": 1,
"best_metric": 0.9994428567862855,
"monitor_metric": "macro_f1",
"early_stopped": true
},
"calibration": {
"method": "task_conditional_sigmoid",
"split": "eval",
"rows": 36000,
"pre_ece": 0.06013730731601039,
"post_ece": 0.0007514102792564575,
"pre_accuracy": 0.9978333333333333,
"post_accuracy": 0.9981666666666666,
"accuracy_delta": 0.0003333333333332966,
"tasks": {
"constraint_scope": {
"rows": 9000,
"pre_ece": 0.0073926387441783925,
"post_ece": 0.00019267901287378653,
"pre_accuracy": 1.0,
"post_accuracy": 1.0,
"accuracy_delta": 0.0
},
"constraint_stability": {
"rows": 3000,
"pre_ece": 0.0035802200536977353,
"post_ece": 0.00016981111181246789,
"pre_accuracy": 1.0,
"post_accuracy": 1.0,
"accuracy_delta": 0.0
},
"constraint_type": {
"rows": 8000,
"pre_ece": 0.0152260080692046,
"post_ece": 0.0002949549691239062,
"pre_accuracy": 1.0,
"post_accuracy": 1.0,
"accuracy_delta": 0.0
},
"fact_type": {
"rows": 6000,
"pre_ece": 0.008209015508071071,
"post_ece": 0.00021106970310058593,
"pre_accuracy": 1.0,
"post_accuracy": 1.0,
"accuracy_delta": 0.0
},
"pii_presence": {
"rows": 10000,
"pre_ece": 0.1278197693399604,
"post_ece": 0.002118116763033487,
"pre_accuracy": 0.9922,
"post_accuracy": 0.9934,
"accuracy_delta": 0.0011999999999999789
}
}
},
"epoch_stats": [
{
"epoch": 1,
"train_loss": 0.07304321517285889,
"train_accuracy": 0.9985243055555556,
"train_macro_f1": 0.9996205356573068,
"train_weighted_f1": 0.9985243053339707,
"valid_loss": 0.07040478728784708,
"valid_accuracy": 0.9978333333333333,
"valid_macro_f1": 0.9994428567862855,
"valid_weighted_f1": 0.9978333319466658,
"monitor_metric": "macro_f1",
"monitor_value": 0.9994428567862855,
"improved": true
},
{
"epoch": 2,
"train_loss": 0.07098693556871623,
"train_accuracy": 0.9985138888888889,
"train_macro_f1": 0.9996178570566361,
"train_weighted_f1": 0.998513888553585,
"valid_loss": 0.06895428509737356,
"valid_accuracy": 0.9979166666666667,
"valid_macro_f1": 0.9994642852803567,
"valid_weighted_f1": 0.9979166649791653,
"monitor_metric": "macro_f1",
"monitor_value": 0.9994642852803567,
"improved": false
},
{
"epoch": 3,
"train_loss": 0.06992293075722475,
"train_accuracy": 0.9984930555555556,
"train_macro_f1": 0.9996124998827812,
"train_weighted_f1": 0.9984930550997047,
"valid_loss": 0.06817925236885357,
"valid_accuracy": 0.9979166666666667,
"valid_macro_f1": 0.9994642852803567,
"valid_weighted_f1": 0.9979166649791653,
"monitor_metric": "macro_f1",
"monitor_value": 0.9994642852803567,
"improved": false
},
{
"epoch": 4,
"train_loss": 0.0692276576162906,
"train_accuracy": 0.9984861111111111,
"train_macro_f1": 0.9996107141570066,
"train_weighted_f1": 0.9984861106105815,
"valid_loss": 0.06766472301272425,
"valid_accuracy": 0.9979166666666667,
"valid_macro_f1": 0.9994642852803567,
"valid_weighted_f1": 0.9979166649791653,
"monitor_metric": "macro_f1",
"monitor_value": 0.9994642852803567,
"improved": false
}
],
"skipped_invalid_tasks": {},
"artifact_scope": "family",
"evaluation_suite": "standard",
"dataset_hashes": {
"train": {
"path": "/nvme/CognitiveMemoryLayer/packages/models/prepared_data/modelpack/extractor_train.parquet",
"sha256": "7001566078ce06fa63779c2025be27b27a1fed4a304133dc334fcbf1336dcd1e",
"bytes": 21540433
},
"test": {
"path": "/nvme/CognitiveMemoryLayer/packages/models/prepared_data/modelpack/extractor_test.parquet",
"sha256": "8fa0af53f8cbd8116335d00659608404ba96055df49b1ab89799418c646a48d3",
"bytes": 2721025
},
"eval": {
"path": "/nvme/CognitiveMemoryLayer/packages/models/prepared_data/modelpack/extractor_eval.parquet",
"sha256": "d066db9c6cadeaea12f4c367262fee444f0992575af2daca06af75f40e9935ce",
"bytes": 2709873
}
}
}