diff --git a/training_configs/llama-3-8b-instruct-simpo-v2.yaml b/training_configs/llama-3-8b-instruct-simpo-v2.yaml index 3d58fa7..77a899f 100644 --- a/training_configs/llama-3-8b-instruct-simpo-v2.yaml +++ b/training_configs/llama-3-8b-instruct-simpo-v2.yaml @@ -31,8 +31,8 @@ max_length: 2048 max_prompt_length: 1800 num_train_epochs: 1 optim: adamw_torch -output_dir: outputs/llama-3-8b-instruct-simpo -run_name: llama-3-8b-instruct-simpo +output_dir: outputs/llama-3-8b-instruct-simpo-v2 +run_name: llama-3-8b-instruct-simpo-v2 per_device_train_batch_size: 2 per_device_eval_batch_size: 4 push_to_hub: false