LoHoSearch is a new benchmark of 544 KG-constructed questions across 11 domains where the strongest search agent scores 34.74% and context strategies add at most 6.8%.
hub Baseline reference
♫ M u S i Q ue: Multihop Questions via Single-hop Question Composition
Baseline reference. 50% of citing Pith papers use this work as a benchmark or comparison.
35
Pith papers citing it
141
external citations · Crossref
Baseline
50% of classified citations
hub tools
citation-role summary
dataset 3
background 2
method 1
citation-polarity summary
years
2026 35representative citing papers
LOCOS scores attention heads via OV-circuit output projection onto answer-token unembedding directions and identifies non-literal retrieval heads whose ablation collapses performance on non-literal benchmarks more than prior literal-copy detectors.
A fixed-iteration spreading activation with per-step cosine similarity gating enables query-aware KG retrieval as one database query, matching QAFD-RAG on MuSiQue while cutting latency.
Prefix gain measured via student-model solve-rate improvement is used to train a Prefix Utility Model (PUM) that supplies stronger supervision than correctness-based process rewards for mathematical reasoning.
LazyAttention kernelizes deferred positional encoding to enable zero-copy, position-agnostic KV cache reuse, delivering 1.37× lower TTFT and 1.40× higher throughput than Block-Attention under skewed document distributions while preserving output quality.
LLM-Wiki structures external knowledge as compilable wiki pages with links and persistent self-correction, achieving SOTA results on HotpotQA, MuSiQue, and 2WikiMultiHopQA by 2.0-8.1 F1 points over prior RAG systems.
EXG is an experience graph framework for self-evolving LLM agents that supports online real-time growth and offline reuse to enhance solution quality and efficiency on code generation and reasoning benchmarks.
F-GRPO factorizes group-relative policy optimization into generation and ranking phases within one autoregressive sequence, using order-invariant coverage and position-aware utility rewards to improve top-ranked performance on recommendation and multi-hop QA tasks.
A VOI-based controller for dual inference budgets improves multi-hop QA performance by prioritizing search actions and selectively finalizing answers.
SIOP enables turn-level credit assignment in LLM agents via semantic clustering of final answers as latent outcomes, improving performance on reasoning benchmarks without verifiers.
ReaLM-Retrieve uses step-level uncertainty to trigger retrievals during reasoning, achieving 10.1% better F1 scores and 47% fewer calls on multi-hop QA benchmarks.
SOB benchmark shows LLMs achieve near-perfect schema compliance but value accuracy of only 83% on text, 67% on images, and 24% on audio.
PASS middleware independently standardizes process/outcome/format streams, derives value-homogeneous chunks, and converts cumulative returns to average value density, yielding consistent pass@1 gains over GRPO baselines in two domains and two signal paradigms.
RSRank learns calibrated relevance scores from alignment between representational shifts induced by candidate documents and those from oracle document sets, enabling zero-threshold filtering.
RASER routers built on one-shot RAG features selectively escalate retrieval, matching SOTA F1 scores on multi-hop QA while using 41-49% of the tokens required by always-prune across six LLMs and three benchmarks.
K-BrowseComp is a new Korean web-browsing agent benchmark where frontier LLMs score 30-46% and Korean LLMs score 0-10% on the verified subset.
Bidirectional Evolutionary Search augments autoregressive expansion with evolutionary recombination operators and dense backward subgoal feedback to produce better candidates than standard best-of-N or tree search for language model self-improvement.
Dynamic Gradient Gating monitors lm_head gradient norms to safely reuse rollout batches in RLVR, achieving up to 2.93x sample efficiency and 2.14x wall-clock speedup across math, ALFWorld, WebShop, and QA tasks.
A²TGPO improves RL policy optimization for multi-turn agentic LLMs by normalizing information gain within same-depth turn groups, rescaling cumulative advantages by sqrt of term count, and modulating clipping ranges per turn's normalized IG.
Attention-based models can retrieve evidence intrinsically by using decoder attention to score and reuse their own pre-encoded chunks, outperforming separate retrieval pipelines on QA benchmarks.
Verbal-R3 uses a verbal reranker to generate analytic narratives that guide retrieval and reasoning in LLMs, achieving SOTA results on complex QA benchmarks.
NeocorRAG uses Evidence Chains to achieve SOTA retrieval quality in RAG on HotpotQA, 2WikiMultiHopQA, MuSiQue, and NQ for 3B and 70B models while using under 20% of the tokens of comparable methods.
Injecting a few vectors near the embedding-space centroid can make them appear in top-k results for up to 94.4% of queries via centrality-driven hubness.
ReCal introduces hierarchical reward decomposition and distribution-aware optimization to address ambiguous credit assignment and optimization bias in RL-based LLM routing.
citing papers explorer
-
LoHoSearch: Benchmarking Long-Horizon Search Agents Beyond the Human Difficulty Ceiling
LoHoSearch is a new benchmark of 544 KG-constructed questions across 11 domains where the strongest search agent scores 34.74% and context strategies add at most 6.8%.
-
Logit-Contribution Scoring Identifies Non-Literal Retrieval Heads
LOCOS scores attention heads via OV-circuit output projection onto answer-token unembedding directions and identifies non-literal retrieval heads whose ablation collapses performance on non-literal benchmarks more than prior literal-copy detectors.
-
Query-Aware Spreading Activation for Multi-Hop Retrieval over Knowledge Graphs
A fixed-iteration spreading activation with per-step cosine similarity gating enables query-aware KG retrieval as one database query, matching QAFD-RAG on MuSiQue while cutting latency.
-
From Correctness to Utility: Gain-Based Prefix Evaluation for LLM Reasoning
Prefix gain measured via student-model solve-rate improvement is used to train a Prefix Utility Model (PUM) that supplies stronger supervision than correctness-based process rewards for mathematical reasoning.
-
LazyAttention: Efficient Retrieval-Augmented Generation with Deferred Positional Encoding
LazyAttention kernelizes deferred positional encoding to enable zero-copy, position-agnostic KV cache reuse, delivering 1.37× lower TTFT and 1.40× higher throughput than Block-Attention under skewed document distributions while preserving output quality.
-
Retrieval as Reasoning: Self-Evolving Agent-Native Retrieval via LLM-Wiki
LLM-Wiki structures external knowledge as compilable wiki pages with links and persistent self-correction, achieving SOTA results on HotpotQA, MuSiQue, and 2WikiMultiHopQA by 2.0-8.1 F1 points over prior RAG systems.
-
EXG: Self-Evolving Agents with Experience Graphs
EXG is an experience graph framework for self-evolving LLM agents that supports online real-time growth and offline reuse to enhance solution quality and efficiency on code generation and reasoning benchmarks.
-
F-GRPO: Factorized Group-Relative Policy Optimization for Unified Candidate Generation and Ranking
F-GRPO factorizes group-relative policy optimization into generation and ranking phases within one autoregressive sequence, using order-invariant coverage and position-aware utility rewards to improve top-ranked performance on recommendation and multi-hop QA tasks.
-
Inference-Time Budget Control for LLM Search Agents
A VOI-based controller for dual inference budgets improves multi-hop QA performance by prioritizing search actions and selectively finalizing answers.
-
Self-Induced Outcome Potential: Turn-Level Credit Assignment for Agents without Verifiers
SIOP enables turn-level credit assignment in LLM agents via semantic clustering of final answers as latent outcomes, improving performance on reasoning benchmarks without verifiers.
-
When to Retrieve During Reasoning: Adaptive Retrieval for Large Reasoning Models
ReaLM-Retrieve uses step-level uncertainty to trigger retrievals during reasoning, achieving 10.1% better F1 scores and 47% fewer calls on multi-hop QA benchmarks.
-
The Structured Output Benchmark: A Multi-Source Benchmark for Evaluating Structured Output Quality in Large Language Models
SOB benchmark shows LLMs achieve near-perfect schema compliance but value accuracy of only 83% on text, 67% on images, and 24% on audio.
-
Process Advantage Signal Shaping: A Paradigm-Agnostic Middleware for Process-Supervised RL in LLM Reasoners
PASS middleware independently standardizes process/outcome/format streams, derives value-homogeneous chunks, and converts cumulative returns to average value density, yielding consistent pass@1 gains over GRPO baselines in two domains and two signal paradigms.
-
RSRank: Learning Relevance from Representational Shifts
RSRank learns calibrated relevance scores from alignment between representational shifts induced by candidate documents and those from oracle document sets, enabling zero-threshold filtering.
-
RASER: Recoverability-Aware Selective Escalation Router for Multi-Hop Question Answering
RASER routers built on one-shot RAG features selectively escalate retrieval, matching SOTA F1 scores on multi-hop QA while using 41-49% of the tokens required by always-prune across six LLMs and three benchmarks.
-
K-BrowseComp: A Web Browsing Agent Benchmark Grounded in Korean Contexts
K-BrowseComp is a new Korean web-browsing agent benchmark where frontier LLMs score 30-46% and Korean LLMs score 0-10% on the verified subset.
-
Self-Improving Language Models with Bidirectional Evolutionary Search
Bidirectional Evolutionary Search augments autoregressive expansion with evolutionary recombination operators and dense backward subgoal feedback to produce better candidates than standard best-of-N or tree search for language model self-improvement.
-
When to Stop Reusing: Dynamic Gradient Gating for Sample-Efficient RLVR
Dynamic Gradient Gating monitors lm_head gradient norms to safely reuse rollout batches in RLVR, achieving up to 2.93x sample efficiency and 2.14x wall-clock speedup across math, ALFWorld, WebShop, and QA tasks.
-
A$^2$TGPO: Agentic Turn-Group Policy Optimization with Adaptive Turn-level Clipping
A²TGPO improves RL policy optimization for multi-turn agentic LLMs by normalizing information gain within same-depth turn groups, rescaling cumulative advantages by sqrt of term count, and modulating clipping ranges per turn's normalized IG.
-
Retrieval from Within: An Intrinsic Capability of Attention-Based Models
Attention-based models can retrieve evidence intrinsically by using decoder attention to score and reuse their own pre-encoded chunks, outperforming separate retrieval pipelines on QA benchmarks.
-
Verbal-R3: Verbal Reranker as the Missing Bridge between Retrieval and Reasoning
Verbal-R3 uses a verbal reranker to generate analytic narratives that guide retrieval and reasoning in LLMs, achieving SOTA results on complex QA benchmarks.
-
NeocorRAG: Less Irrelevant Information, More Explicit Evidence, and More Effective Recall via Evidence Chains
NeocorRAG uses Evidence Chains to achieve SOTA retrieval quality in RAG on HotpotQA, 2WikiMultiHopQA, MuSiQue, and NQ for 3B and 70B models while using under 20% of the tokens of comparable methods.
-
Can You Trust the Vectors in Your Vector Database? Black-Hole Attack from Embedding Space Defects
Injecting a few vectors near the embedding-space centroid can make them appear in top-k results for up to 94.4% of queries via centrality-driven hubness.
-
ReCal: Reward Calibration for RL-based LLM Routing
ReCal introduces hierarchical reward decomposition and distribution-aware optimization to address ambiguous credit assignment and optimization bias in RL-based LLM routing.
-
ProbeScale: Probing Analysis to Optimize Neural Scaling Laws for Efficient Small Language Model Inference
ProbScale finds layer subsets in SLMs like RoBERTa-Large and T5-Base that cut parameters 5-10x while retaining 95-98% of original task performance by maximizing aggregated probe scores under a budget.
-
OCC-RAG: Optimal Cognitive Core for Faithful Question Answering
OCC-RAG develops task-specialized SLMs (0.6B and 1.7B) via a new synthetic data pipeline for multi-hop reasoning and context faithfulness, claiming to match or exceed 2-6x larger general models on HotpotQA, MuSiQue, TAT-QA, ConFiQA, and MuSiQue-Un.
-
SPADER: Step-wise Peer Advantage with Diversity-Aware Exploration Rewards for Multi-Answer Question Answering
SPADER is an RL method for multi-answer QA that claims better recall and F1 via peer-aligned step-level advantages and diversity rewards on four benchmarks.
-
Align Documents to Questions: Question-Oriented Document Rewriting for Retrieval-Augmented Generation
QREAM rewrites documents to question-focused style using iterative ICL and distilled FT models, boosting RAG performance by up to 8% relative improvement.
-
Verify Before You Commit: Towards Faithful Reasoning in LLM Agents via Self-Auditing
SAVeR adds self-auditing of internal beliefs in LLM agents via persona-based candidates and constraint-guided repairs, improving faithfulness on six benchmarks without hurting task performance.
-
Hallucination Basins: A Dynamic Framework for Understanding and Controlling LLM Hallucinations
LLM hallucinations arise from task-dependent basins in latent space, with separability varying by task and geometry-aware steering reducing their probability.
-
Beyond Logical Forms: LLM-Extracted Patterns for Fallacy Classification
LLM-extracted patterns merging logical structures and linguistic cues yield statistically significant gains in fallacy classification over zero-shot baselines with cross-dataset generalization.
-
LongAttnComp: Cross-Family Context Compression for Long-Context Reasoning
LongAttnComp adapts attention-based context compression with token-level chunking, positional reordering, and two-stage fine-tuning to match full-context performance on InfiniteBench Code-Debug and improve multi-document reasoning on LongBench v2 across model families.
-
MMoA: An AI-Agent framework with recurrence for Memoried Mixure-of-Agent
MMoA adds LSTM recurrence to Mixture-of-Agents routing, reaching 58.0% win rate on AlpacaEval 2.0 versus 59.8% for baseline MoA while cutting runtime by up to 4.6%.
-
Optimizing Abstractive Summarization With Fine-Tuned PEGASUS
Fine-tuned PEGASUS achieves state-of-the-art ROUGE scores on XL-Sum English corpus with 4.04% ROUGE-1, 15.25% ROUGE-2, and 3.39% ROUGE-L gains over mT5 baseline.
- LatentSkill: From In-Context Textual Skills to In-Weight Latent Skills for LLM Agents