Open StudyingLover opened 6 months ago
can't we just change the file openai_api.py from 512 to 32 k ?
It seems not work. I have an article , whose token len is 20069, I use tiktoken to compute. I found that if token len more than about 6000, the model will be no respnse(http 200, no output). Is this issue caused from token limit? @x4080
I'm not sure, I just saw that we can chang the limit from that file - Thanks for clarifying that it wont work
same problem
same problem
solved, see #136
我分别量化了chatglm3和chatglm3-32k两个模型,请问如何设置他们的context大小.
我看到很多文件都需要修改,请问能否出一个文档来说明一下