Xu Chen

Real-time robotic inspection can achieve a mean orientation error of only 0.4° by seamlessly integrating human input with perception-driven adjustments.

OneRank achieves superior multi-task recommendation performance by seamlessly integrating task-specific learning within a unified Transformer framework, eliminating the traditional encoder-predictor bottleneck.

Query-aware segment folding boosts long-video understanding performance by up to 9.1 percentage points without additional computational cost.

Achieving near-FP16 accuracy with 4-bit quantization, TwinQuant offers a groundbreaking approach to optimizing large language model inference speed and efficiency.