| { | |
| "dataset": null, | |
| "revision": null, | |
| "source_split": null, | |
| "source_records_scanned": null, | |
| "counts": null, | |
| "corpus_tokens": { | |
| "train": { | |
| "fineweb_edu": 1400000000, | |
| "english_wikipedia": 500000000, | |
| "narrative": 100000000 | |
| }, | |
| "validation": { | |
| "fineweb_edu": 700000, | |
| "english_wikipedia": 250000, | |
| "narrative": 50000 | |
| }, | |
| "test": { | |
| "fineweb_edu": 700000, | |
| "english_wikipedia": 250000, | |
| "narrative": 50000 | |
| } | |
| }, | |
| "split_policy": "Canonical source-document SHA256 buckets: test 0-9, validation 10-19, train 20-999; global passage dedup across splits and Bronze, Silver, and Gold-1", | |
| "filter_version": null, | |
| "source_checkpoint_sha256": "928865f44252b3ff38a23f7ced73f4986e3e3951673be035b824ec15e5ee1306", | |
| "data_manifest_sha256": "cf6ca78377e03aef753b073d25f710d665de31f0a91d285032955feecc565934", | |
| "checkpoint_step": 15259, | |
| "stage_tokens": 2000027648, | |
| "previous_tokens": 3000238080, | |
| "training_tokens": 5000265728, | |
| "sampled_validation_loss": 2.609314309898764, | |
| "export_dtype": "bfloat16", | |
| "optimizer_state_included": false | |
| } | |