owenliang/qwen-vllm

运行vllm_offline.py报错

zjjznw123 opened this issue · 3 comments

zjjznw123 commented

版本cuda12.1 torch 2.1.0 vllm0.2.2,根据你的版本步骤，就报错， self.stop_words_ids=[self.tokenizer.im_start_id,self.tokenizer.im_end_id,self.tokenizer.eos_token_id]
AttributeError: 'Qwen2TokenizerFast' object has no attribute 'im_start_id'，请问是我哪错了

XingYu-Zhong commented

我也遇到同样的问题，请问是哪里错了呀？

rivergao87 commented

请问怎么解决的呢

XingYu-Zhong commented

我用的qwen版本不对，我是qwen1.5的版本