Toward Guidance-Free AR Visual Generation via Condition Contrastive Alignment

Open in new window