Generation config needed to be fixed
#13
by
smart-liu
- opened
In the generation_config.py, the "eos_token_id" is set to 2, but it should be the index of <|im_end|> token. If use other framework to deploy this model, the generation wounld stop.
You are right. Fixed in https://huggingface.co/internlm/internlm2-chat-7b/discussions/12
ZwwWayne
changed discussion status to
closed