Denis Kuznedelev

Hi, @rationalism. We are planning to quantize this model. Since new Mixtral is pretty large, this would take some time. Hopefully, the quantized model will be ready in a week...

Hi, @harshdhamecha. Thanks for your interest in the project. Could you elaborate more on a specific use case? AQLM method is model-agnostic and may be readily applied to common CV...

@LiMa-cas 1) As discussed in the paper, global fine-tuning is a more expressive (and more expensive) finetuning. One optimizes the target loss directly instead of per-block L2 loss, which is...

Hi, @LiMa-cas. Sorry for the delayed reply. The issue seems to be with dataset loading. We have not encountered such an error. Probably it is due to the internet connection....

Hi, @AIpplier. Llama and Qwen are very similar in terms of architecture, and it is likely that the code script working for Llama would work for Qwen without significant changes....

Hi, @LugerW-A we have not tried so far. But I think the language component of the multimodal model should be quantizable in a similar fashion to LLM.