StateLinFormer: Advanced Long-Term Memory for Navigation
Discover how StateLinFormer uses stateful training to enhance long-term memory and improve navigation accuracy in complex environments.
AscendOptimizer: Boost Ascend NPU Operator Performance
Discover how AscendOptimizer enhances Ascend NPU operator efficiency with profiling-in-loop search and kernel optimization for faster performance.
Safe Reinforcement Learning with Preference-Based Constraints
Discover how preference-based constraint inference improves safety in reinforcement learning for complex, real-world decision-making scenarios.
Synthetic Mixed Training: Boosting Language Models Beyond RAG
Discover how Synthetic Mixed Training surpasses RAG by combining synthetic QAs and documents for superior language model performance.
ReCAP: Advanced CAPTCHA Solving for Native GUI Agents
Discover ReCAP, a native GUI agent using automated data and self-corrective training to boost CAPTCHA solving success from 30% to 80%.