ChatGLM2-6B
ChatGLM2-6B copied to clipboard
全参数精调 [launch.py:315:sigkill_handler] Killing subprocess
Is there an existing issue for this?
- [X] I have searched the existing issues
Current Behavior
7张24G的显卡
Expected Behavior
No response
Steps To Reproduce
启动脚本:
Environment
- OS:Ubuntu 18.04.6 LTS
- Python:3.10.9
- Transformers:4.30.2
- PyTorch:2.0.0
- CUDA Support (`python -c "import torch; print(torch.cuda.is_available())"`) : True
Anything else?
No response
你是CUDA out of memory,我把--fp16 改成:
--pre_seq_len 128
--quantization_bit 4
就可以了,但是感觉没有分布式加速,指定了4张gpu,每张卡的现存占用都是一样的,感觉每个gpu都在重复计算,并没有加速
Is there an existing issue for this?
- [x] I have searched the existing issues
Current Behavior
7张24G的显卡 
Expected Behavior
No response
Steps To Reproduce
启动脚本: 
Environment
- OS:Ubuntu 18.04.6 LTS - Python:3.10.9 - Transformers:4.30.2 - PyTorch:2.0.0 - CUDA Support (`python -c "import torch; print(torch.cuda.is_available())"`) : TrueAnything else?
No response
你好,请问这个问题有解决吗?
没有搞定 @xlhuang132
全参数咋调的,求助
7张24G的显卡 
启动脚本: 