| { | |
| "epoch": 30, | |
| "epochs": 30, | |
| "best_val_spearman": 0.6483711214796755, | |
| "steps_per_epoch": 5358, | |
| "total_steps": 160740, | |
| "warmup": 200, | |
| "lr": 5e-05, | |
| "numpy_rng": { | |
| "bit_generator": "PCG64", | |
| "state": { | |
| "state": 227225035480919325237193047808093113659, | |
| "inc": 87136372517582989555478159403783844777 | |
| }, | |
| "has_uint32": 0, | |
| "uinteger": 2723729953 | |
| }, | |
| "fingerprint": { | |
| "student_type": "mlm", | |
| "mlm_checkpoint": "ZurabDz/ka-mlm", | |
| "tokenizer": "ZurabDz/ka-bpe-32k", | |
| "vocab_size": 32000, | |
| "max_len": 64, | |
| "dim": null, | |
| "depth": null, | |
| "heads": null, | |
| "mlp_dim": null, | |
| "embed_dim": null, | |
| "dropout": 0.1, | |
| "batch_size": 84, | |
| "lr": 5e-05, | |
| "weight_decay": 0.01, | |
| "warmup": 200, | |
| "seed": 0, | |
| "val_frac": 0.1, | |
| "center": true, | |
| "reg_weight": 1.0, | |
| "sim_weight": 0.0, | |
| "sim_loss": "kl", | |
| "temperature": 0.05, | |
| "n_sentences": 500000, | |
| "n_train": 450106, | |
| "teacher_dim": 1024, | |
| "sentences_sha256": "4a2f6c9af080bb7d470feebaa76fba29a78c3a47f3b67c808f5f2bc8e0df3d13" | |
| } | |
| } | |