When EOS Tokens Disagree: Understanding Length Inflation in On-Policy Distillation Paper • 2609.20511 • Published 5 days ago • 99
DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression Paper • 2609.19969 • Published 5 days ago • 154
JEPA-Anything: Learning Predictive Models across Different Worlds Paper • 2609.20800 • Published 5 days ago • 59
VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention Paper • 2609.15810 • Published 8 days ago • 49
ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE Tasks Paper • 2609.18805 • Published 6 days ago • 61
Rethinking Critic Learning in PPO: Understanding and Mitigating Value Flattening Paper • 2609.18708 • Published 6 days ago • 75
SAS: Simple Attention Sparsification via End-to-End Optimization of Context Ranking Paper • 2609.13141 • Published 11 days ago • 65
Grouped Value Attention: Efficient KV Caching via On-Demand Key Reconstruction Paper • 2609.13285 • Published 14 days ago • 78