dodecahedron / tokenizer_config.json
bfpill's picture
Upload SmolLM-32M (LlamaForCausalLM, 16K vocab, step 5197, trained on 21.8B tokens of FineWeb)
8925d4e verified
Raw
History Blame Contribute Delete
220 Bytes
{
"backend": "tokenizers",
"bos_token": "<|endoftext|>",
"eos_token": "<|endoftext|>",
"is_local": true,
"model_max_length": 2048,
"pad_token": "<|padding|>",
"tokenizer_class": "PreTrainedTokenizerFast"
}