Add trlx

[cat-state]

Add a link to carperai/trlx, a repo for distributed training of language models with Reinforcement Learning via Human Feedback (RLHF)

[nichtdax]

LGTM