- attn_projector=orthogonal, attn_weight=5, extra_grad_stats=False, learning_rate=0.0002, per_device_train_batch_size=4, warmup_ratio=0
- attn_projector=orthogonal, attn_weight=5, extra_grad_stats=True, learning_rate=0.0002, per_device_train_batch_size=4, warmup_ratio=0
- attn_projector=orthogonal, attn_weight=5, learning_rate=0.0002, per_device_train_batch_size=4, warmup_ratio=0