fix optimizer reset for relora sft (#1414)
* fix optimizer reset * set states to reset for 8bit optimizers and handle quantile runtime error for embeddings * fix relora test to check grad_norm * use flash attn for relora and tweak hyperparams for test * fix messages field for test dataset
This commit is contained in:
@@ -2,3 +2,4 @@ pytest
|
||||
pytest-xdist
|
||||
pytest-retry
|
||||
pytest-sugar
|
||||
tbparse
|
||||
|
||||
Reference in New Issue
Block a user