baichuan2-13b-chat

#1
by trek90s - opened

Hello, may I ask if there is a conversion to the 13B-Chat model? Also, can you provide a script repo for the conversion (do you use llama.cpp, but it doesn't seem to support the model of Baichuan 2)? Thank you very much!

Sorry, I haven't found a good way to convert 13B-Chat. Here's one of my recent failures, https://www.modelscope.cn/models/shaowenchen/baichuan2-13b-chat-gguf. The model's intelligence is behaving abnormally.

No, this repository may work for llama v1 only, not v2. After conversion, it behaves abnormally and the quantized version does not work.

Sorry, I haven't found a good way to convert 13B-Chat. Here's one of my recent failures, https://www.modelscope.cn/models/shaowenchen/baichuan2-13b-chat-gguf. The model's intelligence is behaving abnormally.

Thanks for your relpy, maybe we should wait for the suport of official llama.cpp. Thanks a lot!

llama.cpp upgrades its convert-baichuan-hf-to-gguf.py

Sign up or log in to comment