arXiv:2605. 06772v2 Announce Type: replace Abstract: As large language models (LLMs) show increasing promise on research-level physics reasoning tasks and agentic AI becomes more common, a practical question emerges: How does the interaction between researchers and agents affect the results?
arXiv:2607. 18006v1 Announce Type: cross Abstract: Large language models achieve strong reasoning performance, but often at prohibitive training cost - a challenge that is especially acute for compact models ($\leq 4 \, \mathrm{B}$ parameters) trained under limited budgets.
Large language models achieve strong reasoning performance, but often at prohibitive training cost - a challenge that is especially acute for compact models ($\leq 4 \, \mathrm{B}$ parameters) trained under limited budgets. We introduce MADA-RL, a post-training framework that specializes compact models into generator and critic roles and trains them with a debate-aware learning signal, fine-tuning only a small subset of parameters via LoRA adapters.
arXiv:2610.07335v1 Announce Type: cross Abstract: Improving the reliability of large language model (LLM) agents in long-horizon decision-making remains a key challenge. When deployed as autonomous a...
arXiv:2606. 01667v1 Announce Type: new Abstract: Test-time scaling has become a major way to improve large language model reasoning, but its orchestration has remained designer-engineered: a fixed sample budget, a fixed refinement loop, a fixed scoring rule, or a fixed search policy decides how compute is spent, leaving the model in charge of solving but not of orchestration.
arXiv:2610.09858v1 Announce Type: cross Abstract: Large language model agents tackle multi-step tasks by interleaving reasoning and tool calls with observations from the environment. Prior work has s...
arXiv:2601. 02854v2 Announce Type: replace Abstract: As an agent-level reasoning and coordination paradigm, Multi-Agent Debate (MAD) orchestrates multiple agents through structured debate to improve answer quality and support complex reasoning.
arXiv:2608.30397v1 Announce Type: new Abstract: Natural-language critiques provide supervision beyond scalar rewards for non-verifiable generation, which lacks deterministic verifiers. In critique-gu...
MAMath-Bench is a new benchmark that evaluates multimodal mathematical reasoning in multi‑agent systems (MASs). It contains 1,003 problems across 11 categories, multiple difficulty levels, and modalities such as text, images, and audio, with fine‑grained milestone annotations for correctness, reasoning‑process reliability, and robustness to adversarial perturbations. The study finds that while collaboration can improve intermediate reasoning coverage, it does not consistently boost final‑answer accuracy and can increase redundancy, and current systems remain vulnerable to subtle adversarial errors.
arXiv:2605.26087v2 Announce Type: replace-cross Abstract: Frontier LLMs now perform strongly across a wide range of physics evaluations, but it is hard to disentangle genuine reasoning from recall of...
Seq@K is a new evaluation protocol that measures a language model’s ability to solve a problem within K sequential attempts, receiving feedback after each failure. It isolates the core capability of iterative refinement from feedback, contrasting with existing metrics like Pass@K that do not capture this behavior. Experiments on seven benchmarks and sixteen frontier models show that Seq@K can change model rankings, reveal more efficient configurations, and provide insights into how models use feedback.
Test-time scaling has become a major way to improve large language model reasoning, but its orchestration has remained designer-engineered: a fixed sample budget, a fixed refinement loop, a fixed scoring rule, or a fixed search policy decides how compute is spent, leaving the model in charge of solving but not of orchestration. We introduce ATLAS, an agentic test-time scaling framework in which an LLM orchestrator owns the control loop end-to-end.
The paper introduces the concept of an agent’s "taste"—its ability to make effective long‑horizon decisions—and presents Taste‑Bench, a new benchmark that automatically generates decision‑fork questions from agent trajectories. Taste‑Bench evaluates models on choosing the best path without seeing future outcomes, revealing that top models answer only about 60% of questions correctly and that later‑appearing evidence makes forks harder. The authors also demonstrate that training a student model to mimic a teacher’s judgment improves decision quality and overall success on held‑out software engineering tasks.