Haoyu Wang

KVEraser achieves near-perfect context erasure efficiency with only a fraction of the latency increase compared to traditional methods, revolutionizing how we handle stale information in LLMs.

Acoda can induce LLMs to misinterpret code analysis with a 70% success rate, transforming code security in the age of AI.

UniSVQ achieves state-of-the-art performance in 2-bit quantization, outperforming traditional methods while enhancing inference speed.

LC-QAT achieves superior performance in 2-bit quantization with just a fraction of the training data, setting a new standard for data-efficient model optimization.

Ultra Flash achieves real-time high-resolution video generation at unprecedented frame rates, pushing the boundaries of what’s possible in streaming video AI.

Raw context outperforms compact memory designs, revealing that memory structure is crucial for effective video generation in action-conditioned models.

GMBFormer achieves state-of-the-art urban green-space extraction by leveraging a global memory bank that enhances semantic reuse without compromising on visual fidelity.