Daily Papers of 2026-10-14

1. PhyProbe: Rethinking Physical Consistency Evaluation in Generated Videos 5 upvotes, #1 of 2026-10-14
2. LeWAM: End-to-End World and Action Modeling with JEPAs 0 upvotes, #2 of 2026-10-14
3. DoSA: Dots Sparse Attention 0 upvotes, #2 of 2026-10-14
4. Harness^2: Recursive Agent Harnessing for an Open World 0 upvotes, #2 of 2026-10-14
5. Vela: User Behavior as Supervision Sparks LLM Intelligence 0 upvotes, #2 of 2026-10-14
6. ShellElastic: Scalable Elastic Pretraining with Simultaneous Maximal Updates and Hyperparameter Transfer 0 upvotes, #2 of 2026-10-14
7. PANINI-DR: Controlled Deep Research through Generative Semantic Workspaces 0 upvotes, #2 of 2026-10-14
8. RSIPO: Recursive Self-Improving Policy Optimization for Agentic Reinforcement Learning via Skill-Policy Co-Evolution 0 upvotes, #2 of 2026-10-14

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.