[msp26]

>This works well for me except the 15B+ don't run fast enough on a 4090

I assume quantized models will run a lot better. TheBloke already seems like he's on it.

https://huggingface.co/TheBloke/CodeLlama-13B-fp16

[thewataccount]

Unfortunately what I tested was StarCoder 4bit. We really need exllama which should make even 30b viable from what I can tell.

Because codellama is llama based it may just work possibly?