• Constrained Harness Editing Delivers Limited but Transferable Gains. RRSI gained up to 14.1 points on its evolution split and 4.7 points across five out-of-distribution benchmarks. It also consumed 30% fewer policy tokens.

• WorldCrafter Compresses History for the Target Camera View. It turns multiview observations into a fixed number of view-specific tokens without explicit depth-based correspondence. The paper reports better long-term consistency and camera control during minute-long exploration.

• Token-Level Correction May Cut Annotation Costs While Preserving the Model Distribution. onPanda keeps most of the model’s own output. A small controlled study found a 52% reduction in median annotation time versus human post-editing.

Also Notable

• GameHorizon Combines Offline Scores With Stepwise In-Game Failures — Its multiscale evaluation covers 21 games, 5,000 hours of expert footage, and more than one million model calls..
• Jev-Mem Adds a Lightweight Control Plane for Memory — It manages memory routing, budget allocation, and stopping decisions. The system calls a slower model only for complex reasoning and reports both quality and latency gains on LoCoMo..
• Harness-Zero Distills Harness Behavior Into Model Weights — After removing specialized harnesses, it still raised macro-average task success from 23.3% to 44.3%..
• D-RAC Cuts Chunking Output Tokens by 95.7% — The comparison uses agentic chunking with frontier LLMs. D-RAC converts pages into retrieval-friendly Markdown, then plans IDs for deterministic text units..
• VideoGen-Agent Improves When Given Better Generation Tools — Replacing its generator raised its VABench score from 75.6 to 86.1 without retraining. The agent coordinates enhancement, generation, and verification tools..
• ACLArena Tracks Forgetting Across Multistage Post-Training — It analyzes forgetting at model and token levels across multi-teacher on-policy distillation, self-distillation fine-tuning, and model merging. Its method combines offline replay of high-quality trajectories with routed LoRA experts specialized through RL..
• Complex KDA Represents 2D Rotation With One Structured Update — It uses a single diagonal-plus-rank-one update. The method only expands the allowed gate and delta values while preserving non-expansiveness and the original computational structure..
• KernelBench’s Official Checker Misses 16.9% of Independently Confirmed Failures — The result comes from 10,303 compilable faults injected into 188 problems. The miss rate reaches 78.6% for numerical-accuracy faults..
• EvidenT Verifies Retrieved Evidence Before Answering — It uses deterministic lexical alignment. Across about 500 real enterprise queries, it increased matches to gold-standard sources by 29% on average versus a prompting baseline..
• ARPS Reaches 87.3% Success on Zero-Adaptation LIBERO-Plus Transfer — Its action expert reads only compact, future-facing, action-relevant predicted states. It outperformed Fast-WAM by 39.2 percentage points..
• Similar Recommendation-Change Rates Do Not Determine the Authority Gap — AuthEval scores model recommendations separately from the final choices made by local services. The rate alone cannot determine the gap’s size or direction..

Read the full edition →