swap batch size for gradient accumulation steps to decouple from num gpu
This commit is contained in:
@@ -22,7 +22,7 @@ wandb_watch:
|
||||
wandb_run_id:
|
||||
wandb_log_model:
|
||||
output_dir: ./stable-alpaca-3b
|
||||
batch_size: 2
|
||||
gradient_accumulation_steps: 1
|
||||
micro_batch_size: 1
|
||||
num_epochs: 1
|
||||
optimizer: adamw_bnb_8bit
|
||||
|
||||
Reference in New Issue
Block a user