Isn't LoRA solved problem by unsloth?
Unsloth doesn't support distributed training well and doesn't support Kimi models.
Distributed training has been on their to do list for a good long while iirc
Model compression more generally is far from a solved field
Unsloth doesn't support distributed training well and doesn't support Kimi models.
Distributed training has been on their to do list for a good long while iirc
Model compression more generally is far from a solved field