mistral-community/mixtral-8x22B-Instruct-v0.3-original
#121
by
yttria
- opened
8x22B Instruct v0.1 quants are broken since b3028 (Tokenizer WPM fixes), please try v0.3.
sure, it's queued, should be done... not sure when, the mixtral 8x22b's are a bit fiddly, so might take a few days.
mradermacher
changed discussion status to
closed
ah, and, please provide a url to the hf model next time, thanks
unfortunately, the only model with that name on hf is not a transformers model and is not supported by llama.cpp.