sf_medium_bf16 / tokenizer_config.json
kastan's picture
First attempt at a BitsNBytes QLoRA Falcon model.
aa1f0cc
raw
history blame
180 Bytes
{
"add_prefix_space": false,
"clean_up_tokenization_spaces": true,
"eos_token": "<|endoftext|>",
"model_max_length": 2048,
"tokenizer_class": "PreTrainedTokenizerFast"
}