TanaroSch
TanaroSch
@Inserian I encounter the same error on ubuntu 20.04 with nvidia-hpc-sdk module enabled. Do you know if there might be another error preventing llama from using nccl?
> can you make the same change for litellm.acompletion on main > > (otherwise our ci/cd will fail tests) Done. Also fixed other occurrences of max_tokens with float type. Edit:...
@ChristianPejrup I think the issue is only with embedding models. I managed to deploy R1 and Phi-4 as well, using the same /models endpoint without the error messages. I just...