Rethinking Token Reweighting for SFT: Suppress, Reverse, and Extrapolate Learned Features Paper • 2609.33463 • Published 9 days ago • 11
Sample Count Is Not Enough: Candidate-Generation Strategy Shapes the Energy and Performance of LLM Test-Time Scaling Paper • 2609.19499 • Published 20 days ago • 37
Mask Forcing: Improving Autoregressive Video Diffusion Distillation via Dual-Noise Masking Rollout Paper • 2609.09123 • Published 28 days ago • 55