Jing Liu

Qwen-VLA introduces a unified embodied foundation model that extends vision-language understanding to continuous action generation, enabling robust, multi-task generalization across diverse robotic tasks and embodiments.

The paper introduces Cookie-Bench, a novel, autonomous, and reference-free evaluation framework that significantly improves the assessment of interactive web generation capabilities for frontier LLMs.

RACE-Sched is an asynchronous agentic framework that successfully integrates low-latency, real-time scheduling decisions with advanced, long-horizon reasoning provided by Large Language Models.

WaveFilter is a novel, training-free framework that uses wavelet transforms to efficiently filter critical tokens in the KV cache, significantly improving the long-context performance of Diffusion LLMs.

This paper proposes Head-Chunked Multi-Stream Pipeline (HCMS) to exploit the computational independence of multi-head attention and achieve fine-grained communication-computation overlap, resulting in speedups of up to 17.5% over the Ulysses baseline.

The paper proposes CAPTAIN, a perplexity-based APT detector using pre-trained language models with minimal preprocessing.