| { | |
| "base": "Qwen/Qwen3.5-9B", | |
| "method": "fsdp2-custom-loop", | |
| "n_rows": 22321, | |
| "n_self": 0, | |
| "n_sequences": 11081, | |
| "n_tokens": 17506727, | |
| "world": 4, | |
| "micro_batch": 4, | |
| "accum": 4, | |
| "steps": 173, | |
| "lr": 1e-05, | |
| "optim": "AdamW fused, fp32 master, bf16 compute, fp32 grad reduce", | |
| "loss_curve": [ | |
| 1.403472900390625, | |
| 1.2008408308029175, | |
| 1.2321603298187256, | |
| 1.2248389720916748, | |
| 1.193364143371582, | |
| 1.1675400733947754, | |
| 1.2074540853500366, | |
| 1.0775381326675415, | |
| 1.184540033340454, | |
| 1.1193952560424805, | |
| 1.1095657348632812, | |
| 1.0941717624664307, | |
| 1.1998660564422607, | |
| 1.1657809019088745, | |
| 1.1528652906417847, | |
| 1.1288039684295654, | |
| 1.1920957565307617, | |
| 1.1281521320343018, | |
| 1.1178914308547974, | |
| 1.134727954864502, | |
| 1.0643160343170166, | |
| 1.1739487648010254, | |
| 1.1533012390136719, | |
| 1.198366641998291, | |
| 1.2317473888397217, | |
| 1.1055439710617065, | |
| 1.1493632793426514, | |
| 1.100700855255127, | |
| 1.1283423900604248, | |
| 1.1580134630203247, | |
| 1.1361677646636963, | |
| 1.2649364471435547, | |
| 1.1410850286483765, | |
| 1.1824312210083008, | |
| 1.0584875345230103 | |
| ], | |
| "loss_first": [ | |
| 1.403472900390625 | |
| ], | |
| "loss_last": [ | |
| 1.0584875345230103 | |
| ], | |
| "train_seconds": 2077.6351487636566, | |
| "tokens_per_second": 8426.275908172698, | |
| "resumed_from": null | |
| } |