ID and Graph View Contrastive Learning with Multi-View Attention Fusion for Sequential Recommendation — ThinkLLM