Model training details
#3395
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
{{ message }}
-
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?
Beta Was this translation helpful? Give feedback.
All reactions