Copernicus-Tokenizer / tokenizer_config.json
Nj-1111's picture
Upload folder using huggingface_hub
6cba08c verified
Raw
History Blame
593 Bytes
{
"tokenizer_class": "PreTrainedTokenizerFast",
"model_max_length": 4096,
"bos_token": "<|endoftext|>",
"eos_token": "<|endoftext|>",
"unk_token": "<|unk|>",
"pad_token": "<|pad|>",
"additional_special_tokens": [
"<think>",
"</think>",
"<scratchpad>",
"</scratchpad>",
"<verify>",
"</verify>",
"<reflect>",
"</reflect>",
"<|user|>",
"<|assistant|>",
"<|system|>",
"<|tool_call|>",
"<|tool_result|>",
"<|sep|>",
"<|im_start|>",
"<|im_end|>"
],
"clean_up_tokenization_spaces": false,
"add_prefix_space": false
}