16 lines
743 B
JSON
16 lines
743 B
JSON
{
|
|
"backend": "tokenizers",
|
|
"bos_token": "<|begin_of_text|>",
|
|
"clean_up_tokenization_spaces": true,
|
|
"eos_token": "<|eot_id|>",
|
|
"legacy": false,
|
|
"model_input_names": [
|
|
"input_ids",
|
|
"attention_mask"
|
|
],
|
|
"model_max_length": 131072,
|
|
"pad_token": "<|finetune_right_pad_id|>",
|
|
"padding_side": "left",
|
|
"tokenizer_class": "PreTrainedTokenizerFast",
|
|
"chat_template": "{{- bos_token }}{%- for message in messages %}{%- set content = '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n' + message['content'] | trim + '<|eot_id|>' %}{%- if loop.index0 == 0 %}{%- set content = bos_token + content %}{%- endif %}{{ content }}{%- endfor %}{{ '<|start_header_id|>assistant<|end_header_id|>\n\n' }}"
|
|
} |