The model output is garbled.
#1
by zhouyizoe - opened
Hello, I used the sft_noact.pth weights you provided and performed inference using the dataset you released. However, the result was garbled characters. Could you please tell me where the problem lies? I did not modify the released code.
zhouyizoe changed discussion status to closed
zhouyizoe changed discussion status to open
I found that the base model I used was incorrect. After replacing it, the output became correct.