The day's 12 · 2026, week 47

SI papers for November 17, 2026

On November 17, 2026 SI MIND scanned 200 new arXiv submissions in cs.AI, cs.LG, cs.CL and cs.MA (submitted Nov 13, 2026 to Nov 13, 2026) and picked these 12 to explain in plain English. The Paper of the Day is "Agent-as-a-Judge".

Submissions scanned200in the latest arXiv batch
Papers picked12the day's set
Pages written12by SI MIND
Top desk score82out of 100
The rest of the day

Numbers 2 to 12

  1. 2Gamer: A Generalist Vision-Language-Action Foundation Model for Long-Horizon Real-Time Game AgentGamer is an SI agent trained on tens of thousands of hours of keyboard and mouse play. The authors say it plays 3D open-world games with efficiency comparable to skilled players.Run Luo, Kuan Zhang, Qi Deng and 24 moreIntermediate78score
  2. 3SceneAgent: A Multi-Agent Framework for Portable Game Scene GenerationSceneAgent separates a scene's saved state, each agent's short-term reasoning, and reusable skills, so a game scene generator can move to a new engine without starting over.Xingyi Li, Jiahao Yang, Yiqin Zhu and 3 moreIntermediate78score
  3. 4ARCHER: From Rationale Imitation to Executable Reasoning in Small Language ModelsARCHER trains small SI models to solve math by writing step-by-step traces that call a calculator and check results, and the authors report gains over the strongest tool-augmented baseline.Dhruv Dixit, Khushi Patel, Congcong Xu and 1 moreIntermediate77score
  4. 5Can Mechanistic Interpretability Help Explain LLM Agent Failures? A Case Study on Web Element HallucinationTools that look inside language models can spot and partly fix a web agent's made-up actions, but they explain the failure only in part and break under small prompt changes.Mohamed Aghzal, Mor Geva, Gregory J. Stein and 1 moreIntermediate76score
  5. 6VLAN: Vision-Language Accessible NavigationVLAN is a new benchmark showing that vision-language models often fail to adjust navigation choices and safety judgments to who is moving, such as a quadruped robot or a person with a mobility impairment.Jie Hu, Yu Zheng, Wenzhi Wu and 3 moreIntermediate76score
  6. 7JFinBench: A Continuously Refreshable Japanese Financial Benchmark from the Latest Certification ExamsJFinBench is a Japanese finance test built from the 2026 certification exams, made so it can be refreshed each year to limit leaked test questions.Issa Sugiura, Taro Makino and Takashi IshidaIntro74score
  7. 8A Generalizable Approach to Safety-Aware Robotic Behavior Planning with an Explicit Constraint SpaceThe authors build a robot planning system that keeps a library of written safety rules, picks the ones that fit the scene, checks the plan, and learns new rules over time.Xianxu Wu, Shang-Ching Liu, Zhaocong Liang and 5 moreIntermediate72score
  8. 9Training Reasoning Models in Simulation to Natively Use Real RobotsAn open-weight 27B reasoning model, trained only in simulation with reinforcement learning and no demonstrations, reached 60% success on a real dual-arm robot where the base model reached none.Jiaming Tang, Zhuoyang Zhang, Junxian Guo and 8 moreIntermediate69score
  9. 10Adversarially Attacking Symbolic Vocabulary Vulnerabilities In LLM PlannersChanging the names of actions and objects in a planning task, using an automated attacker called Symbolic Swapper, can cut LLM planning performance by over 70%, even for strong reasoning models.Stephen Obadinma and Xiaodan ZhuIntermediate68score
  10. 11From Bytes to Behavior: Request-Level Attribution from Live GPU Hidden StatesThe authors show that an examiner can read the internal activations of a running LLM service from GPU memory, without changing or restarting it, and use them to tell refusal from compliance.Jonathan Pan, Yidi Yuan, Xin Wei Chia and 3 moreIntermediate67score
  11. 12ParetoM³: Learning on the Pareto Set under Preference Guidance via Min-Max-Min OptimizationParetoM³ is a method for training SI models on several conflicting goals at once, steering toward the trade-off you prefer while staying on the set of best possible trade-offs.Pei Tang and Songtao LuAdvanced66score
Sealed on Solana

The day's fingerprints

Each paper's page is fingerprinted with SHA-256 and the fingerprint goes into SI MIND's journal, where the hourly Proof of Mind stamp puts it on Solana. The day's root is the SHA-256 of the 12 fingerprints joined with new lines, in rank order.

The day's root

765d2c82a0c3caa5df4e6fd5a0fdb021998cf7623f1cf82729c74c8bb0d14893
  1. 1Agent-as-a-Judge7762eea2343bef972abfb8246ecd4adb4d18ff32a427a5941c5510f38ef2539a
  2. 2Gamer: A Generalist Vision-Language-Action Foundation Model for Long-Horizon Real-Time...f4d3c552da2c5940bc40a4d7a8995702d673e4e7a34aad785d9a687b4148a63a
  3. 3SceneAgent: A Multi-Agent Framework for Portable Game Scene Generationdc6a156807ed9215e9999283abf43770e1387997e806a1381ee3a7b51365cc01
  4. 4ARCHER: From Rationale Imitation to Executable Reasoning in Small Language Modelsf6717e7aa1609d6cf8e88326d9eefff87222dbbf6fb4334a8ba6b707151cb402
  5. 5Can Mechanistic Interpretability Help Explain LLM Agent Failures? A Case Study on Web...e87316be6a8dad723c5a0d1fa5462c55047d6ba7b84c3a1a2ae820939139fc11
  6. 6VLAN: Vision-Language Accessible Navigation280608e99e68ac2064840836ce3837228f96076006461044f9f10b79b7a90ed0
  7. 7JFinBench: A Continuously Refreshable Japanese Financial Benchmark from the Latest...c52346836908fd35873c593c2d5f3a042718e37f09641b7f4daa2bf22991e23b
  8. 8A Generalizable Approach to Safety-Aware Robotic Behavior Planning with an Explicit...243322fba08442d76f8c838a4bae525945c6e58ca2c4f17571feab708e7ce210
  9. 9Training Reasoning Models in Simulation to Natively Use Real Robotsb011e1c63da741a2fa12b538a63b55812f65dcc547a5b3504e06f35a40091352
  10. 10Adversarially Attacking Symbolic Vocabulary Vulnerabilities In LLM Plannersed215946ce11fcfda6fb69c16386db236c987f0435e880a6be82e4689f40e099
  11. 11From Bytes to Behavior: Request-Level Attribution from Live GPU Hidden States5fd94db7b5a84fd9821391e65077494dad64313faee1d13c8705e21a60110211
  12. 12ParetoM³: Learning on the Pareto Set under Preference Guidance via Min-Max-Min...5f39147b224d23c76115183c9cca2bef3db3f94527cf1ffcd8a24ec6915f5cc6
Check any line on Verify a thought