Upload artifacts/stage4_run_metadata.json with huggingface_hub
Browse files
artifacts/stage4_run_metadata.json
ADDED
|
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"stage": 4,
|
| 3 |
+
"dataset_id": "THemidli/applied-ner-stage4-final",
|
| 4 |
+
"base_model": "prajjwal1/bert-mini",
|
| 5 |
+
"base_revision": "5e123abc2480f0c4b4cac186d3b3f09299c258fc",
|
| 6 |
+
"tokenizer_model": "google/bert_uncased_L-2_H-128_A-2",
|
| 7 |
+
"tokenizer_revision": "30b0a37ccaaa32f332884b96992754e246e48c5f",
|
| 8 |
+
"seed": 20260802,
|
| 9 |
+
"labels": [
|
| 10 |
+
"O",
|
| 11 |
+
"B-PERSON",
|
| 12 |
+
"I-PERSON",
|
| 13 |
+
"B-ORGANIZATION",
|
| 14 |
+
"I-ORGANIZATION",
|
| 15 |
+
"B-LOCATION",
|
| 16 |
+
"I-LOCATION",
|
| 17 |
+
"B-TIMEDATE",
|
| 18 |
+
"I-TIMEDATE",
|
| 19 |
+
"B-PRODUCT",
|
| 20 |
+
"I-PRODUCT",
|
| 21 |
+
"B-WORKOFART",
|
| 22 |
+
"I-WORKOFART",
|
| 23 |
+
"B-JOB",
|
| 24 |
+
"I-JOB",
|
| 25 |
+
"B-AMOUNT",
|
| 26 |
+
"I-AMOUNT"
|
| 27 |
+
],
|
| 28 |
+
"parameter_count": 11109137,
|
| 29 |
+
"trainable_parameter_count": 11109137,
|
| 30 |
+
"device": "mps",
|
| 31 |
+
"platform": "macOS-27.0-arm64-arm-64bit",
|
| 32 |
+
"torch_version": "2.13.0",
|
| 33 |
+
"train_records": 841,
|
| 34 |
+
"test_records": 159,
|
| 35 |
+
"train_chunks": 865,
|
| 36 |
+
"test_chunks": 165,
|
| 37 |
+
"max_length": 256,
|
| 38 |
+
"wall_seconds": 32.908,
|
| 39 |
+
"trainer_metrics": {
|
| 40 |
+
"train_runtime": 32.8519,
|
| 41 |
+
"train_samples_per_second": 421.285,
|
| 42 |
+
"train_steps_per_second": 13.637,
|
| 43 |
+
"total_flos": 49534075078794.0,
|
| 44 |
+
"train_loss": 0.23767771824662173,
|
| 45 |
+
"epoch": 16.0
|
| 46 |
+
},
|
| 47 |
+
"hyperparameters": {
|
| 48 |
+
"epochs": 16,
|
| 49 |
+
"learning_rate": 0.0005,
|
| 50 |
+
"train_batch_size": 32,
|
| 51 |
+
"eval_batch_size": 64,
|
| 52 |
+
"weight_decay": 0.02,
|
| 53 |
+
"label_smoothing_factor": 0.0,
|
| 54 |
+
"warmup_steps": 45,
|
| 55 |
+
"scheduler": "linear",
|
| 56 |
+
"hidden_dropout": 0.1,
|
| 57 |
+
"attention_dropout": 0.1,
|
| 58 |
+
"classifier_dropout": 0.1
|
| 59 |
+
},
|
| 60 |
+
"train_overall": {
|
| 61 |
+
"overall_precision": 0.9988776655443322,
|
| 62 |
+
"overall_recall": 0.998653500897666,
|
| 63 |
+
"overall_f1": 0.9987655706430254,
|
| 64 |
+
"overall_accuracy": 0.9998373013720918
|
| 65 |
+
},
|
| 66 |
+
"test_overall": {
|
| 67 |
+
"overall_precision": 0.618925831202046,
|
| 68 |
+
"overall_recall": 0.6747211895910781,
|
| 69 |
+
"overall_f1": 0.6456202756780791,
|
| 70 |
+
"overall_accuracy": 0.8791745256851214
|
| 71 |
+
},
|
| 72 |
+
"test_f1_change_vs_stage3": 0.1771147178353455,
|
| 73 |
+
"mps_run_variance_note": "On the MPS backend, repeated runs with identical seed and config showed ±0.01–0.015 F1 variation (0.6331 vs 0.6455 across the two recorded runs); the seed is fixed, and the variation does not change the model ranking (both runs far above BERT-Tiny, slightly below ELECTRA-Small).",
|
| 74 |
+
"dataset_revision": "ce231175a828d21865c638d9718c4a6e8ba1fb1c",
|
| 75 |
+
"vocab_sha256": "07eced375cec144d27c900241f3e339478dec958f92fddbc551f295c992038a3"
|
| 76 |
+
}
|