03/19/2025 16:24:18 - INFO - root - args: Namespace(output_name='llama1b_lora', datasets=['harmful_behavior'], student_model='distilbert/distilgpt2', teacher_model='distilbert/distilgpt2', alpha=0.0, temperature=1.0, num_train_epochs=1, learning_rate=0.001, max_num_tokens=1024, batch_size=1, grad_acc_steps=16, weight_decay=0.01, adam_epsilon=1e-08, warmup_steps=0, max_grad_norm=1.0, dropout=0.1, kl_loss_weight=0, exclude_neg=True, no_weights=True, optim='adamw_torch', max_steps=1, lr_scheduler_type='linear', logging_steps=50, save_epochs=10, seed=2, resume_from_checkpoint=False, hub_strategy='all_checkpoints', report_to='none', push_to_hub=True, proportions=[1.0], split='train', streaming=True, sequence_length=512, model_dir='./trained/', is_lora_model=False, lora=True, r=16, lora_alpha=32, lora_dropout=0.1, task_type='CAUSAL_LM', output_dir='./trained/llama1b_lora', logger=)