Files
babylm-2026-fr-92m/tokenizer_config.json

9 lines
204 B
JSON
Raw Permalink Normal View History

{
"tokenizer_class": "PreTrainedTokenizerFast",
"model_max_length": 512,
"bos_token": "<|endoftext|>",
"eos_token": "<|endoftext|>",
"unk_token": "<|endoftext|>",
"pad_token": "<|padding|>"
}