Model training details #3395

Unanswered

flodorner asked this question in Q&A

flodorner
Jun 12, 2023

Is there public information about how the oasst llama models were trained?

In particular, is oasst-rlhf-2-llama-30b-7k-steps-xor derived from oasst-sft-7-llama-30b-xor and if so, what hyperparameters were used for rlhf?

Replies: 0 comments

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment