@@ -434,6 +434,7 @@ eval_batch_size: 2
num_epochs: 3
warmup_steps: 100
learning_rate: 0.00003
lr_quadratic_warmup:
logging_steps:
save_steps: # leave empty to save at each epoch
eval_steps:
The note is not visible to the blocked user.