gpt2-sft / tokenizer_config.json
Jennny's picture
Training in progress, step 100
5d5ba09 verified
raw
history blame
680 Bytes
{
"add_prefix_space": false,
"added_tokens_decoder": {
"50256": {
"content": "<|endoftext|>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": true
}
},
"bos_token": "<|endoftext|>",
"chat_template": "{% for dialogue in chosen %}\n{% if dialogue.speaker == 'Human' %}Human: {{ dialogue.text + eos_token }}{% else %}Assistant: {{ dialogue.text + eos_token }}{% endif %}\n{% endfor %}",
"clean_up_tokenization_spaces": true,
"eos_token": "<|endoftext|>",
"model_max_length": 1024,
"pad_token": "<|endoftext|>",
"tokenizer_class": "GPT2Tokenizer",
"unk_token": "<|endoftext|>"
}