| { |
| "model": "OpenSML-Base-175M", |
| "pretraining_step": 132005, |
| "processed_token_positions": 17300111360, |
| "weight_sha256": "5151567e266244c9e32fd6e01775192633abe227320269ecdefe28ce43e9ccc3", |
| "validation": { |
| "val_loss": 2.5563359551876785, |
| "val_ppl": 12.888506628383553, |
| "source_losses": { |
| "web": 2.941684268414974, |
| "dclm": 2.9943208396434784, |
| "wiki": 1.987750370055437, |
| "cosmopedia": 1.7873524278402328, |
| "python": 1.2724680323153734 |
| }, |
| "validation_fingerprint": "4ddd461c131d4d27d07bb07f5aab3fcb5a00d4b1202066e3d9b6ee6b3129087b", |
| "batches_per_source": 32, |
| "batch_size": 2, |
| "sequence_length": 2048, |
| "sources": 5, |
| "nominal_token_positions": 655360, |
| "scope": "Historical fixed document validation, mixture-weighted loss. Not a downstream benchmark or a fresh test." |
| }, |
| "downstream_benchmarks": {}, |
| "scope": "No JSONLite post-training or extraction metrics are attributed to these base weights." |
| } |
|
|
