Replies: 2 comments
-
第141行的 eval() 替换为quantized(8) |
Beta Was this translation helpful? Give feedback.
-
In client.py
change into ↓
line 151 |
Beta Was this translation helpful? Give feedback.
-
第141行的 eval() 替换为quantized(8) |
Beta Was this translation helpful? Give feedback.
-
In client.py
change into ↓
line 151 |
Beta Was this translation helpful? Give feedback.
-
您好!
我的显存不足,只有10GB,因此想要运行INT8精度的模型,我发现在ChatGLM3/composite_demo/client.py文件下有如下代码似乎可以开启量化:
注释部分有提示 # plus .quantized() if you want to use quantized model,本人纯小白一枚,请问大神.quantized()我加在哪里能开启量化?
Beta Was this translation helpful? Give feedback.
All reactions