Model, filtered dataset, code, evaluations and paper for 'Constraining Language Modeling Pretraining Data'
e8fb262 verified | { | |
| "vocab_size": 2700, | |
| "tokens_per_story": 150.4, | |
| "tokens_per_word": 1.211, | |
| "roundtrip_exact": "1412/1412", | |
| "lexicon_forms_single_token": 0.7967 | |
| } |