training acceleration

5 papers

• Counteractive RL: Rethinking Core Principles for Efficient and Scalable Deep Reinforcement Learning
• Curriculum Design for Trajectory-Constrained Agent: Compressing Chain-of-Thought Tokens in LLMs
• DuSA: Fast and Accurate Dual-Stage Sparse Attention Mechanism Accelerating Both Training and Inference
• Improving Energy Natural Gradient Descent through Woodbury, Momentum, and Randomization
• Stable Coresets via Posterior Sampling: Aligning Induced and Full Loss Landscapes