estuday

Results 12 comments of estuday

> 参考https://github.com/fastnlp/CPT/tree/master/finetune/generation 十分感谢

> 多了逗号和空格 去掉空格和逗号后依旧是空的。。请问怎么办

> 看一下issue94在 2023年3月24日,14:00,estuday ***@***.***> 写道: 多了逗号和空格 去掉空格和逗号后依旧是空的。。请问怎么办 —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***> ![image](https://user-images.githubusercontent.com/45581281/227453905-9443b95d-0480-4950-9b7c-5e25f4cfb4a9.png) ![image](https://user-images.githubusercontent.com/45581281/227453962-e2f8bc01-3cab-433a-87ee-c103d9b800be.png) 都试过了,都是为空

> 我也是类似的问题: ![1679907965562](https://user-images.githubusercontent.com/44674527/227895507-33d3ff40-7cf1-4d18-87eb-8abc1bc26a89.png) 感觉程序卡死在了某个地方 下载最新的有报错,我已经解决了,在代码里加代理代理ip和端口就行

i also find another question,after the openllm start command is run, if you press ctrl+c to end the process, port 3000 is always occupied and needs to be released manually

> I haven't used torchrun with this code. Does it work with just `python ./main_accelerate.py ` ? The same problem occurred just use `python ./main_accelerate.py `, can you show the...

补充完成一轮训练后的输出情况: ![image](https://github.com/QwenLM/Qwen/assets/45581281/5aaa05ca-00f8-449a-850b-50c34a7d81f2)

> How did you conduct the inference? model.chat(query,tokenizer,history)

> Please first try adjusting the repetition penalty (higher), the temperature (higher), and the top_p (higher) in the generation_config.json. Hi, I adjusted these parameters and the effect seemed to be...