When reflections localize early errors, in-context search solves exp-small pass-rate problems with poly sequential attempts; otherwise it offers no asymptotic gain over parallel sampling, and the update is learnable and RLVR-optimal.
hub
Advances in neural information processing systems , volume=
38 Pith papers cite this work. Polarity classification is still indexing.
38
Pith papers citing it
hub tools
citation-role summary
background 2
citation-polarity summary
years
2026 38roles
background 2polarities
background 2representative citing papers
AgentODE uses LLMs to discover ODE structures and infer parameter distributions from aggregate data, recovering consistent structures on benchmarks and RDEB clinical data with 231 observations from 46 patients.
Fork-think with confidence identifies forking points via model confidence in a single path before sampling continuations, cutting tokens up to 30% and runtime up to 57% on reasoning benchmarks while matching or exceeding parallel thinking performance.
A GRPO-trained rubric generator that injects step-level criteria into ReAct agents improves deep-research benchmark scores over several test-time compute baselines.
DISA decouples partition function estimation using offline importance sampling for distribution-matching LLM-RL, matching or exceeding online baselines like FlowRL on math and code benchmarks while retaining more strategy diversity.
LLM-generated combinatorial solvers achieve highest correctness when the model formalizes problems for verified backends rather than attempting to optimize search, which often causes regressions.
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
NoisyCausal benchmark tests LLMs on causal reasoning with structured noise, and a modular LLM-plus-causal-graph framework outperforms baselines while generalizing to Cladder.
OptiVerse is a new benchmark spanning neglected optimization domains that shows LLMs suffer sharp accuracy drops on hard problems due to modeling and logic errors, with a Dual-View Auditor Agent proposed to improve performance.
A reward model trained on 1,130 human preference judgments outperforms base agents by 120 Elo points on a 50-task benchmark for recovering from harmful states in computer-use environments.
Validity-calibrated reasoning distillation improves transfer of reasoning skills by modulating updates based on relative local validity of next steps instead of enforcing full trajectory imitation.
GraphBit is a DAG-based engine-orchestrated framework for agentic LLMs that achieves 67.6% accuracy with zero hallucinations on GAIA benchmarks.
Stale-occupancy sequential fine-tuning of multi-agent LLMs incurs an O(n²) certificate penalty; TeamTR resamples under intermediate occupancy and enforces token-level trust regions to restore O(n) scaling and stable gains.
InduceKV is a retrieval-based continual adaptation method that uses bilevel selection to build a compact set of inducing KV memories for fixed-footprint updates to multimodal LLMs.
SEAR trains one LLM via adversarial process rewards to explore harmful reasoning paths but flip to safe outputs, reducing over-refusal while preserving safety.
CLExEval introduces a human-annotated evaluation framework on 40 rare cases that identifies verbosity bias, hidden knowledge paradox, and 68.6% reasoning-to-output mismatch in LLMs while showing LLM-as-a-Judge overestimates reliability.
ISM maintains a self-refined bank of verified strategy schemas to enable continual mathematical reasoning in frozen LLMs, outperforming baselines on MATH-Hard and OlympiadBench while using 64-86% fewer schemas.
DynaSteer is a dynamic representation editing framework that uses pattern clustering, Fisher-LDA, and lookahead entropy monitoring to steer LLM reasoning trajectories toward truth on MATH and coding tasks.
High-Entropy Sum (HES) selects high-quality reasoning data for LLMs by summing entropy of the top highest-entropy tokens, matching full-dataset performance with top 20% in SFT and outperforming baselines in RFT and RL.
SSOPD converts intra-group correct-wrong contrast into process supervision by distilling a teacher distribution from the shortest correct completion into prefixes of the longest wrong completion, improving GRPO on AIME and HMMT benchmarks.
Partial harnesses for LLM agents, specifying only initial execution steps, achieve higher pass rates than fully decomposed workflows, as analyzed through trajectory alignment and validated in synthetic and terminal benchmarks.
Nexa learns a response-conditioned policy that starts with parallel agent execution and adds at most one round of sequential message passing via a predicted sparse DAG, strictly subsuming pure parallel mode.
Many-shot CoT-ICL improves when demonstrations are ordered for smooth conceptual progression, with CDS delivering up to 5.42 percentage-point gains on math tasks using 64 examples.
Expanded recall in LLM agents erodes cooperative intent in multi-agent social dilemmas, observed in 18 of 28 model-game settings.
citing papers explorer
-
When Does In-Context Search Help? A Sampling-Complexity Theory of Reflection-Driven Reasoning
When reflections localize early errors, in-context search solves exp-small pass-rate problems with poly sequential attempts; otherwise it offers no asymptotic gain over parallel sampling, and the update is learnable and RLVR-optimal.
-
LLM-Guided ODE Discovery and Parameter Inference from Small-Cohort Aggregate Data
AgentODE uses LLMs to discover ODE structures and infer parameter distributions from aggregate data, recovering consistent structures on benchmarks and RDEB clinical data with 231 observations from 46 patients.
-
Fork-Think with Confidence
Fork-think with confidence identifies forking points via model confidence in a single path before sampling continuations, cutting tokens up to 30% and runtime up to 57% on reasoning benchmarks while matching or exceeding parallel thinking performance.
-
Co-ReAct: Rubrics as Step-Level Collaborators for ReAct Agents
A GRPO-trained rubric generator that injects step-level criteria into ReAct agents improves deep-research benchmark scores over several test-time compute baselines.
-
DISA: Offline Importance Sampling for Distribution-Matching LLM-RL
DISA decouples partition function estimation using offline importance sampling for distribution-matching LLM-RL, matching or exceeding online baselines like FlowRL on math and code benchmarks while retaining more strategy diversity.
-
Formalize, Don't Optimize: The Heuristic Trap in LLM-Generated Combinatorial Solvers
LLM-generated combinatorial solvers achieve highest correctness when the model formalizes problems for verified backends rather than attempting to optimize search, which often causes regressions.
-
Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
-
NoisyCausal: A Benchmark for Evaluating Causal Reasoning Under Structured Noise
NoisyCausal benchmark tests LLMs on causal reasoning with structured noise, and a modular LLM-plus-causal-graph framework outperforms baselines while generalizing to Cladder.
-
OptiVerse: A Comprehensive Benchmark towards Optimization Problem Solving
OptiVerse is a new benchmark spanning neglected optimization domains that shows LLMs suffer sharp accuracy drops on hard problems due to modeling and logic errors, with a Dual-View Auditor Agent proposed to improve performance.
-
Human-Guided Harm Recovery for Computer Use Agents
A reward model trained on 1,130 human preference judgments outperforms base agents by 120 Elo points on a 50-task benchmark for recovering from harmful states in computer-use environments.
-
Validity-Calibrated Reasoning Distillation
Validity-calibrated reasoning distillation improves transfer of reasoning skills by modulating updates based on relative local validity of next steps instead of enforcing full trajectory imitation.
-
GraphBit: A Graph-based Agentic Framework for Non-Linear Agent Orchestration
GraphBit is a DAG-based engine-orchestrated framework for agentic LLMs that achieves 67.6% accuracy with zero hallucinations on GAIA benchmarks.
-
TeamTR: Trust-Region Fine-Tuning for Multi-Agent LLM Coordination
Stale-occupancy sequential fine-tuning of multi-agent LLMs incurs an O(n²) certificate penalty; TeamTR resamples under intermediate occupancy and enforces token-level trust regions to restore O(n) scaling and stable gains.
-
InduceKV: Fixed-Footprint Continual Adaptation of Multimodal LLMs via Inducing KV Memories
InduceKV is a retrieval-based continual adaptation method that uses bilevel selection to build a compact set of inducing KV memories for fixed-footprint updates to multimodal LLMs.
-
Addressing Over-Refusal in LLMs with Competing Rewards
SEAR trains one LLM via adversarial process rewards to explore harmful reasoning paths but flip to safe outputs, reducing over-refusal while preserving safety.
-
CLExEval: A Human-in-the-Loop Framework for Qualitative Evaluation of LLM Clinical Reasoning
CLExEval introduces a human-annotated evaluation framework on 40 rare cases that identifies verbosity bias, hidden knowledge paradox, and 68.6% reasoning-to-output mismatch in LLMs while showing LLM-as-a-Judge overestimates reliability.
-
ISM:Self-Improving Strategy Memory for Continual Mathematical Reasoning
ISM maintains a self-refined bank of verified strategy schemas to enable continual mathematical reasoning in frozen LLMs, outperforming baselines on MATH-Hard and OlympiadBench while using 64-86% fewer schemas.
-
Search for Truth from Reasoning: A Dynamic Representation Editing Framework for Steering LLM Trajectories
DynaSteer is a dynamic representation editing framework that uses pattern clustering, Fisher-LDA, and lookahead entropy monitoring to steer LLM reasoning trajectories toward truth on MATH and coding tasks.
-
Unified Data Selection for LLM Reasoning
High-Entropy Sum (HES) selects high-quality reasoning data for LLMs by summing entropy of the top highest-entropy tokens, matching full-dataset performance with top 20% in SFT and outperforming baselines in RFT and RL.
-
Self-Supervised On-Policy Distillation for Reasoning Language Models
SSOPD converts intra-group correct-wrong contrast into process supervision by distilling a teacher distribution from the shortest correct completion into prefixes of the longest wrong completion, improving GRPO on AIME and HMMT benchmarks.
-
Harnesses for Inference-Time Alignment over Execution Trajectories
Partial harnesses for LLM agents, specifying only initial execution steps, achieve higher pass rates than fully decomposed workflows, as analyzed through trajectory alignment and validated in synthetic and terminal benchmarks.
-
Response-Conditioned Parallel-to-Sequential Orchestration for Multi-Agent Systems
Nexa learns a response-conditioned policy that starts with parallel agent execution and adds at most one round of sequential message passing via a predicted sparse DAG, strictly subsuming pure parallel mode.
-
Many-Shot CoT-ICL: Making In-Context Learning Truly Learn
Many-shot CoT-ICL improves when demonstrations are ordered for smooth conceptual progression, with CDS delivering up to 5.42 percentage-point gains on math tasks using 64 examples.
-
The Memory Curse: How Expanded Recall Erodes Cooperative Intent in LLM Agents
Expanded recall in LLM agents erodes cooperative intent in multi-agent social dilemmas, observed in 18 of 28 model-game settings.
-
PaT: Planning-after-Trial for Efficient Test-Time Code Generation
PaT defers planning until after failed trials in LLM code generation, enabling heterogeneous cheap-plus-powerful model setups that match large-model performance at roughly 69% lower cost.
-
ReFlect: An Effective Harness System for Complex Long-Horizon LLM Reasoning
ReFlect is a harness that wraps LLMs to detect and recover from reasoning errors, achieving 7-29 pp gains over direct CoT on long-horizon tasks and improving code patch quality to 82-87%.
-
Power Distribution Bridges Sampling, Self-Reward RL, and Self-Distillation
The power distribution is the target of power sampling, the closed-form solution to self-reward KL-regularized RL, and the basis for power self-distillation that matches sampling performance at lower cost.
-
FlexSQL: Flexible Exploration and Execution Make Better Text-to-SQL Agents
FlexSQL reaches 65.4% on Spider2-Snow by allowing agents to flexibly explore schemas, generate diverse plans, choose SQL or Python execution, and apply two-tiered repair.
-
Beyond Individual Mimicry: Constructing Human-Like Social network with Graph-Augmented LLM Agents
GraphMind equips LLM agents with graph awareness to construct human-like social networks, producing botnets that substantially degrade performance of both text-based and graph-based detectors.
-
When Web Agents Finish but Still Fail: Reproducible Triggers and Trace Diagnostics for Parallel Web Exploration
Parallel WebBench reveals GRPO training raises web agent completion to 96% but leaves a large correctness gap from context-bound loops, premature termination, and synthesis collapse.
-
From Agent Traces to Trust: A Survey of Evidence Tracing and Execution Provenance in LLM Agents
This survey defines execution provenance as a typed graph of agent execution and evidence tracing as its projection onto evidence-support relations, then reviews methods, taxonomy, benchmarks, and challenges for auditable LLM agents.
-
RADAR: Redundancy-Aware Diffusion for Multi-Agent Communication Structure Generation
RADAR generates query-adaptive multi-agent communication structures via conditional discrete graph diffusion guided by effective graph size, outperforming baselines on accuracy and token consumption across six benchmarks.
-
Can LLMs Take Retrieved Information with a Grain of Salt?
LLMs exhibit systematic failures in obeying expressed certainty in retrieved contexts, but a combination of prior reminders, certainty recalibration, and context simplification reduces obedience errors by 25%.
-
BALAR : A Bayesian Agentic Loop for Active Reasoning
BALAR is a task-agnostic Bayesian loop that maintains structured beliefs over latent states, selects questions via expected mutual information, and expands its state space when needed, delivering 14.6-38.5% accuracy gains over baselines on detective, puzzle, and clinical diagnosis benchmarks.
-
Dual-Cluster Memory Agent: Resolving Multi-Paradigm Ambiguity in Optimization Problem Solving
DCM-Agent improves LLM optimization solving by 11–21% on seven benchmarks via dual-cluster memory of Approaches, Checklists, and Pitfalls plus adaptive path switching.
-
KnowledgeBerg: Evaluating Systematic Knowledge Coverage and Compositional Reasoning in Large Language Models
KnowledgeBerg is a 4,800-question, 17-language benchmark showing LLMs fail at systematically enumerating bounded knowledge universes and performing compositional set-based reasoning over them.
-
DORA Explorer: Improving the Exploration Ability of LLMs Without Training
DORA Explorer boosts LLM agent exploration without training by ranking diverse actions using log-probabilities and a tunable parameter, yielding UCB-competitive results on multi-armed bandits and gains on text adventure environments.
-
SGR: A Stepwise Reasoning Framework for LLMs with External Subgraph Generation
SGR enhances LLM reasoning accuracy by generating external subgraphs from knowledge bases and guiding progressive inference over them, yielding consistent gains over baselines on benchmarks.