mradermacher

/

model_requests

like 58

How I love gated models. Anyway, it's queued and should be done in a few hours or so. Cheers!

Thank you thank you thank you, much love!

Unfortunately, it turns out to not be supported by llama.cpp at this time (due to lack of pretokenizer support). Sorry :/

This should now be supported in latest llama.cpp (b3542). I ran convert_hf_to_gguf.py and performed Q5_K_M quantization myself and everything worked perfectly fine.

Let's give it another try then :)