Model, filtered dataset, code, evaluations and paper for 'Constraining Language Modeling Pretraining Data'
e8fb262 verified | { | |
| "train": { | |
| "source_stories": 2718842, | |
| "kept": 282603 | |
| }, | |
| "validation": { | |
| "source_stories": 13792, | |
| "kept": 1412 | |
| }, | |
| "k": 300, | |
| "max_unk": 8, | |
| "source": "karpathy/tinystories-gpt4-clean", | |
| "invalid_dropped": 8536, | |
| "lexicon": { | |
| "lemmas": 300, | |
| "forms": 900, | |
| "names": 50 | |
| } | |
| } |