Text Generation
Transformers
Safetensors
llama
UNA
juanako
cybertron
xaberius
Eval Results
Inference Endpoints
text-generation-inference

Reproduce using 24/48GB VRAM?

#7
by gameveloster - opened

Is 24GB VRAM enough to perform this finetune? Or 48GB VRAM?

@chargoddard can advise better, but platy-34b which is the father was FT on a 6000's series.
Definitively reducing its precision around the training session will affect the ultimate result.

Make sure u get the latest tokenizer.

fblgit changed discussion status to closed

Sign up or log in to comment