Can you share the finetuning settings
#23
by
Forceless
- opened
Thanks for this great work for
@zai-org-3
!
I've been trying to fine-tune this model lately, but noticed the model seems to lack robustness during both training and inference.
I could only find the batch size and sequence length mentioned in the paper, but other important hyperparameters (e.g., learning rate) seem to be missing.
It would be very helpful for the community and me to use, thx!
Forceless
changed discussion title from
Can you share the learning settings
to Can you share the finetuning settings