what commit was used to quanitze this?
#1
by
belisarius
- opened
what commit was used to quanitze this?
Yes, Im also following the pull request to add the model.
Did you use the latest commit from there? (e5fe089210116acc64d3938884d52f0d0088822f)
Sorry, I've actually used https://github.com/ngxson/llama.cpp/tree/xsn/hunyuan-moe
I'll might need to re-quantize once the final verson will be merged into llama.cpp master.
Till than I'll add the branch link to the card