Daily Papers of 2026-09-17

  1. ScienceIDE: Turning World's Scientific Codebase into Agent Learnable Environments 100 upvotes, #1 of 2026-09-17
  2. Rethinking Critic Learning in PPO: Understanding and Mitigating Value Flattening 79 upvotes, #2 of 2026-09-17
  3. Confidence Comes from Experience: Experiential Confidence Estimation from Reasoning to Agents 71 upvotes, #3 of 2026-09-17
  4. ProgramDistill: From Interactive Web Apps to Verifiable Reference-Guided SWE Tasks 62 upvotes, #4 of 2026-09-17
  5. Agora: Git as Shared Memory for Collective AutoResearch 57 upvotes, #5 of 2026-09-17
  6. VC-Attention: Value Smoothing and Softmax Casting for Low-bit Attention 50 upvotes, #6 of 2026-09-17
  7. ActionPiece: Rethinking Action Tokenization for Autoregressive Vision-Language-Action Models 47 upvotes, #7 of 2026-09-17
  8. Zing-0.5: Toward Playable Worlds with Real-Time Joint Action and Text Control 44 upvotes, #8 of 2026-09-17
  9. EvolveTrade: Experience-Driven Policy Refinement for Self-Evolving LLM Trading Agents 39 upvotes, #9 of 2026-09-17
  10. SpectralShift: Effective Context Window Extension of Gated DeltaNet via Spectral Reparameterization 33 upvotes, #10 of 2026-09-17
  11. HypoEvolve: Genetic Algorithms Enable Multi-Agent LLMs to Discover Scientific Hypotheses 30 upvotes, #11 of 2026-09-17
  12. Gaze as Evidence for Common Grounding: A Cross-Corpus Analysis of MapTask and MUNDEX 30 upvotes, #11 of 2026-09-17
  13. A Zeroth-Order Paradigm for LLM Preference Alignment 29 upvotes, #13 of 2026-09-17
  14. EventEgoHands++: Event-based Egocentric 3D Hand Mesh Reconstruction with Real Dataset 28 upvotes, #14 of 2026-09-17
  15. In-Context Robot Learning with VLM Agents 22 upvotes, #15 of 2026-09-17
  16. Flattening Every Memory Peak in Long-Context Mixture-of-Experts Training 19 upvotes, #16 of 2026-09-17
  17. PANORAMA: Panoptic Grounded Captioning via Mask Proposal Selection 17 upvotes, #17 of 2026-09-17
  18. The Other Half of the Memory Wall: Serving 35B MoEs from SSD with Trained Routing Prediction 16 upvotes, #18 of 2026-09-17
  19. CERA-MoA: Co-Evolving Routing Mechanisms with Continually Learning LLM Agents 14 upvotes, #19 of 2026-09-17
  20. Fathom: Per-Query Read Depth for Sparse Decoding over Offloaded KV Caches 6 upvotes, #20 of 2026-09-17
  21. Assessing nnU-Net Generalization across Brain Tumor Populations in BraTS-GoAT 2026 3 upvotes, #21 of 2026-09-17
  22. Fingers as Legs: Learning Self-Supported Locomotion and Manipulation with an Anthropomorphic Hand 3 upvotes, #21 of 2026-09-17

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.