@@ -571,7 +571,7 @@ torch_compile_backend: # Optional[str]
# training hyperparameters
gradient_accumulation_steps: 1
micro_batch_size: 2
eval_batch_size: 2
eval_batch_size:
num_epochs: 3
warmup_steps: 100
learning_rate: 0.00003
The note is not visible to the blocked user.