arxiv arXiv cs.AI · 56d ago · research

RTSGameBench: An RTS Benchmark for Strategic Reasoning

from English

RTSGameBench addresses limitations in existing RTS benchmarks by offering diverse gameplay, targeted competency diagnosis, and self-evolving scenario generation. It evaluates vision-language models in strategic reasoning under uncertainty, revealing that state-of-the-art models struggle with multiagent coordination and large-scale tasks.

Importance 3/3 New feature vs. leaders New harness with differentiators arXiv cs.AI OpenAI Anthropic Google DeepMind AI agents Multimodal Reasoning models

Read original

arxiv arXiv cs.AI · 55d ago · 2 views

Dual-Agent Framework for Cross-Model Verified Translation

A dual-agent framework converts natural-language experiment protocols into executable commands for robotic lab platforms. It uses a Parser Agent and a rule-based mapping engine to translate protocols, with a heterogeneous LLM Validation Agent ensuring accuracy and triggering self-correction. The framework successfully enables end-to-end autonomous execution of microplate-based experiments like the Bradford assay.

arxiv arXiv cs.AI · 56d ago · 1 view

OneCanvas: 3D Scene Understanding via Panoramic Reprojection

OneCanvas enables 3D scene understanding in Vision-Language Models by aggregating patch features onto a panoramic canvas using 3D world coordinates. It achieves state-of-the-art results on SQA3D and VSI-Bench, with strong generalization on SPBench, using significantly less training compute than prior methods.

arxiv arXiv cs.CL · 56d ago · 2 views

Turing-RL: Learning User Simulators with Turing Rewards

Turing-RL introduces a reinforcement learning method using an LLM judge to evaluate how indistinguishable generated responses are from real user inputs. It outperforms baseline methods in both LLM and human evaluations across chat and Reddit forum domains, demonstrating that optimizing for indistinguishability improves user simulator performance.

arxiv arXiv cs.LG · 56d ago

ViGOS: Decoupling Perception and Reasoning in Multimodal On-Policy Self-Distillation

ViGOS introduces a visually grounded on-policy self-distillation framework for multimodal large language models. It decouples perception and reasoning by using an image-only teacher for visual descriptions and a reasoning teacher for final outputs, reducing reliance on text-only references. This approach improves image-grounded performance across multiple vision-language benchmarks.

arxiv arXiv cs.CL · 56d ago · 1 view

Decoupling Search from Reasoning in LLM Agents

Decoupled Search Grounding (DSG) separates search functionality from reasoning models, enabling vendor-agnostic, tunable, and reusable search grounding. DSG achieves near-native accuracy on SimpleQA with 91% lower search cost and 99.4% warm-cache hit rate, while reducing latency by 68% and preserving concise output contracts.