fix github issue for quantize

#49
Knowledge Engineering Group (KEG) & Data Mining at Tsinghua University org
No description provided.
Knowledge Engineering Group (KEG) & Data Mining at Tsinghua University org

fix github issue of #1059 #1052

zRzRzRzRzRzRzR changed pull request status to merged

新版本 量化文件,似乎优化掉了一个功能。

早先的版本是支持: gpu 无法一次性载入全部,但是CPU 可以一次载入全部时,可以用cpu 载入全部,量化后,gpu进行推理。

https://huggingface.co/THUDM/chatglm3-6b/discussions/47#663cb3c8a4d7c8c9038c5312

Sign up or log in to comment