HF has been pretty good for training in my experience. Both accelerate, deepspeed and other improvements are available and the ecosystem is vibrant. One thing y

HF has been pretty good for training in my experience. Both accelerate, deepspeed and other improvements are available and the ecosystem is vibrant. One thing you might want to look into is to write your own training loop rather than using Autotrainer, but that's if you want to get you hands dirty and get finer control over how you call layers, your loss function etc. Inference is where it falls short really, and solutions like vLLM are much much faster.