TRIAGE evaluates LLMs on prospective metacognitive control by requiring a single plan for task selection, sequencing, and token allocation under a calibrated budget, revealing substantial gaps in current models across math, science, code, and knowledge tasks.
hub Mixed citations
The danger of overthinking: Examining the reasoning-action dilemma in agentic tasks
Mixed citation behavior. Most common role is background (67%).
16
Pith papers citing it
Background
67% of classified citations
hub tools
citation-role summary
background 5
other 1
citation-polarity summary
representative citing papers
AcademiClaw is a new benchmark of 80 student-sourced academic tasks where the best frontier AI agents achieve only a 55% pass rate.
SADU benchmark shows top VLMs reach only 70% accuracy on software architecture diagram tasks, revealing gaps in visual reasoning for engineering artifacts.
Dynamic Rollout Editing reduces overthinking in RL-trained LLMs by editing post-answer continuations in successful rollouts and preferring the edited versions within GRPO groups.
Exploratory interview study with 17 developers identifies four forms of emergent oversight work for software agents and documents situated challenges and heuristics.
Stopping large reasoning models at the first correct reasoning prefix improves accuracy up to 21% by avoiding harmful overthinking that destabilizes correct trajectories.
A hierarchical genetic algorithm induces overthinking in black-box large reasoning models by perturbing logical structure, achieving up to 26.1x longer outputs on the MATH benchmark.
ICR creates a virtual shorter distribution from shortest correct on-policy responses to regularize RL post-training toward concise yet accurate reasoning, improving the accuracy-length Pareto frontier on math and knowledge benchmarks.
FitText embeds evolutionary retrieval of tool descriptions into the agent loop, yielding 2.7-10.6 point NDCG@5 gains on ToolRet and 26.7-point pass-rate gains on StableToolBench.
TRACE aggregates answer consistency and confidence trajectory over multiple reasoning steps to decide when to halt inference, reducing token usage by 25-30% while keeping accuracy within 1-2% of full reasoning.
APMPO boosts average Pass@1 scores on math reasoning benchmarks by 3 points over GRPO by using an adaptive power-mean policy objective and feedback-driven clipping bounds in RLVR training.
FREIA applies free energy principles and adaptive advantage shaping to unsupervised RL, outperforming baselines by 0.5-3.5 Pass@1 points on math reasoning with a 1.5B model.
DTSR enables large reasoning models to dynamically assess chain-of-thought sufficiency via reflection signals and a sufficiency check, reducing reasoning length by 28.9-34.9% with minimal performance loss on Qwen3 models.
Introduces PAS and FAS task abstractions plus the LLM-S^3 benchmark to evaluate LLMs on generating sociodemographic survey responses across 11 real datasets and multiple models.
Self-aligned reward uses relative perplexity differences to encourage concise, query-specific reasoning in LLMs, yielding 4% accuracy gains and 30% lower inference cost when added to PPO or GRPO.
The paper unifies perspectives on Long CoT in reasoning LLMs by introducing a taxonomy, detailing characteristics of deep reasoning and reflection, and discussing emergence phenomena and future directions.
citing papers explorer
-
TRIAGE: Evaluating Prospective Metacognitive Control in LLMs under Resource Constraints
TRIAGE evaluates LLMs on prospective metacognitive control by requiring a single plan for task selection, sequencing, and token allocation under a calibrated budget, revealing substantial gaps in current models across math, science, code, and knowledge tasks.