Blur

Results 7 issues of Blur

### Is there an existing issue for this? - [X] I have searched the existing issues ### Current Behavior 新的checkpoint,网页可以打开,但是无返回结果,部署代码如下: origin_checkpoint = '/devops/chatglm-6b' ptuning_checkpoint = '/devops/20230412/checkpoint-50000' config = AutoConfig.from_pretrained(origin_checkpoint, trust_remote_code=True,...

我的执行命令:CUDA_VISIBLE_DEVICES=1 python3 /mnt/cpfs/guanyutao/Chinese-Vicuna/generate.py \ --model_path /mnt/cpfs/guanyutao/model_checkpoint/llama-7b-hf \ --lora_path /mnt/cpfs/guanyutao/model_checkpoint/Vicuna_finetune/checkpoint-10000 \ --use_local 1 \ --use_typewriter 1 执行运行到这个地方就没有反应了,也没有报错,就是卡在这里。我理解gradio应该输出一个网址。 ![image](https://user-images.githubusercontent.com/59723064/233817207-41399ebf-a670-4da8-900f-22c4ed16867c.png)

模型是baichuan-7B 同样的问题,py脚本中do_predict的回复质量明显高于web_demo

用train_sft的do_predict预测了200条,没有"Human: ",但baichuan微调后的web demo回复会出现"Human: " ![企业微信截图_16872280027975](https://github.com/hiyouga/LLaMA-Efficient-Tuning/assets/59723064/8e6115bd-8686-4857-8a9c-40dac45e6915) ![企业微信截图_16872280403898](https://github.com/hiyouga/LLaMA-Efficient-Tuning/assets/59723064/c6763bb4-8219-4e54-b47a-61fe13d7729f)

### 是否已有关于该错误的issue或讨论? | Is there an existing issue / discussion for this? - [X] 我已经搜索过已有的issues和讨论 | I have searched the existing issues / discussions ### 该问题是否在FAQ中有解答? | Is there an...

使用firefly-baichuan-13b 微调得到firefly-baichuan-13b-sft,微调后合并firefly-baichuan-13b-sft 与Baichuan-13B-Base 进行推理,报错: ![image](https://github.com/yangjianxin1/Firefly/assets/59723064/e2c30151-7c64-4f0e-8042-b07b588553a9)

### Is there an existing issue for this? - [X] I have searched the existing issues ### Current Behavior 之前用的好好的,今天下午网址就打不开了,web_demo.py也不出来gradio的网址了。 ### Expected Behavior _No response_ ### Steps To Reproduce 1...