A new benchmark reveals that language models including GPT-3 are truthful on only 58% of questions designed to elicit popular misconceptions, far below human performance of 94%, with larger models performing worse.
super hub Canonical reference
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Canonical reference. 84% of citing Pith papers cite this work as background.
199
Pith papers citing it
14
external citations · Pith
Background
84% of classified citations
abstract
Large pre-trained language models have been shown to store factual knowledge in their parameters, and achieve state-of-the-art results when fine-tuned on downstream NLP tasks. However, their ability to access and precisely manipulate knowledge is still limited, and hence on knowledge-intensive tasks, their performance lags behind task-specific architectures. Additionally, providing provenance for their decisions and updating their world knowledge remain open research problems. Pre-trained models with a differentiable access mechanism to explicit non-parametric memory can overcome this issue, but have so far been only investigated for extractive downstream tasks. We explore a general-purpose fine-tuning recipe for retrieval-augmented generation (RAG) -- models which combine pre-trained parametric and non-parametric memory for language generation. We introduce RAG models where the parametric memory is a pre-trained seq2seq model and the non-parametric memory is a dense vector index of Wikipedia, accessed with a pre-trained neural retriever. We compare two RAG formulations, one which conditions on the same retrieved passages across the whole generated sequence, the other can use different passages per token. We fine-tune and evaluate our models on a wide range of knowledge-intensive NLP tasks and set the state-of-the-art on three open domain QA tasks, outperforming parametric seq2seq models and task-specific retrieve-and-extract architectures. For language generation tasks, we find that RAG models generate more specific, diverse and factual language than a state-of-the-art parametric-only seq2seq baseline.
hub tools
citation-role summary
background 34
method 3
other 1
citation-polarity summary
claims ledger
- abstract Large pre-trained language models have been shown to store factual knowledge in their parameters, and achieve state-of-the-art results when fine-tuned on downstream NLP tasks. However, their ability to access and precisely manipulate knowledge is still limited, and hence on knowledge-intensive tasks, their performance lags behind task-specific architectures. Additionally, providing provenance for their decisions and updating their world knowledge remain open research problems. Pre-trained models with a differentiable access mechanism to explicit non-parametric memory can overcome this issue, b
authors
co-cited works
representative citing papers
GPT-3 shows that scaling an autoregressive language model to 175 billion parameters enables strong few-shot performance across diverse NLP tasks via in-context prompting without fine-tuning.
Self-GC governs agent context as indexed objects with planner-proposed actions, achieving 84.85% no-impact on future continuations on a hard set versus 54-70% for baselines.
A causal audit of LMLMs finds near-zero parametric leakage after deletion, with surviving correctness coming from retrieval artifacts in the database.
SOLAR is a learning-augmented policy for semantic cache replacement that achieves constant competitive ratio 3 and 5-75% gains over FIFO on retrieval workloads.
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.
UCOB uses local return comparisons between skill and no-skill views to choose which view teaches the other, improving agent training on ALFWorld and WebShop.
LibEvoBench benchmark shows LLMs are version-oblivious on evolving APIs, with documentation helping but version specification not.
PreAct compiles successful agent executions into verifiable state-machine programs for 8.5-13x faster replay on repeated tasks, with an independent evaluator check before storing each program.
MetaSyn is a stage-level benchmark of 442 meta-analyses showing LLM agents retrieve up to 90.9% of eligible studies but include at most 52.7% in their final reports.
DCPM reorganizes LLM agent memory into a cognitive hierarchy driven by a synchronous daytime belief writer and an asynchronous nighttime schema engine, reporting gains on cross-session inference benchmarks.
Re-ranking retrieval candidates via a cross-encoder trained on continuous perturbation-based attribution scores improves citation faithfulness and gold-answer alignment in legal QA over semantic similarity.
Tool schema compression by 44-50% enables agentic RAG at 8K context where uncompressed schemas fail, with +20.5 pp exact match lift across models and scaling to over 800 tools.
Spectral Retrieval uses multi-scale sinc convolutions on token embeddings to interpolate between per-token MaxSim and mean-pooling, achieving large gains on synthetic and LIMIT-small benchmarks for localized retrieval.
IdioLink introduces a benchmark dataset and evaluation showing that strong embedding models struggle to retrieve equivalent meanings across idiomatic and literal forms, relying on shallow cues instead.
PluRule is a new multimodal multilingual benchmark showing that state-of-the-art vision-language models perform only marginally better than a trivial baseline at detecting specific rule violations in pluralistic online communities.
Memory for long-horizon agents should preserve distinctions that affect decisions under a fixed budget, not descriptive features, yielding an exact forgetting boundary and a new online learner DeMem with regret guarantees.
MemFlow routes queries by intent to tiered memory operations, nearly doubling accuracy of a 1.7B SLM on long-horizon benchmarks compared to full-context baselines.
ROME generates deceptive safety benchmarks that degrade LLM agent judgment performance, while ARISE uses analogical retrieval to improve safety decisions at inference time without retraining.
In 30-step recursive LLM loops, append-mode persistent escape from source basins reaches 50% near 400 tokens under full history but plateaus below 50% under tail-clip memory policy, while replace-mode switching largely reflects state reset.
OCR-Memory encodes agent trajectories as images with visual anchors and retrieves verbatim text via locate-and-transcribe, yielding gains on long-horizon benchmarks under strict context limits.
A transformer trained on random meaningless MicroPy programs generalizes to execute diverse human-written programs, providing empirical evidence it can act as a universal computer.
XGRAG uses graph perturbations to quantify component contributions in GraphRAG and achieves 14.81% better explanation quality than text-based baselines on QA datasets, with correlations to graph centrality.
SUIN improves CTR prediction by augmenting target user sequences with similar users' behaviors via embedding-based retrieval, user-specific position encoding, and user-aware target attention.
citing papers explorer
-
TruthfulQA: Measuring How Models Mimic Human Falsehoods
A new benchmark reveals that language models including GPT-3 are truthful on only 58% of questions designed to elicit popular misconceptions, far below human performance of 94%, with larger models performing worse.
-
Language Models are Few-Shot Learners
GPT-3 shows that scaling an autoregressive language model to 175 billion parameters enables strong few-shot performance across diverse NLP tasks via in-context prompting without fine-tuning.
-
Self-GC: Self-Governing Context for Long-Horizon LLM Agents
Self-GC governs agent context as indexed objects with planner-proposed actions, achieving 84.85% no-impact on future continuations on a hard set versus 54-70% for baselines.
-
Auditing Forgetting in Limited Memory Language Models
A causal audit of LMLMs finds near-zero parametric leakage after deletion, with surviving correctness coming from retrieval artifacts in the database.
-
When Classic Cache Policies Fail: Learning-Augmented Replacement for Semantic Retrieval Buffers
SOLAR is a learning-augmented policy for semantic cache replacement that achieves constant competitive ratio 3 and 5-75% gains over FIFO on retrieval workloads.
-
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.
-
UCOB: Learning to Utilize and Evolve Agentic Skills via Credit-Aware On-Policy Bidirectional Self-Distillation
UCOB uses local return comparisons between skill and no-skill views to choose which view teaches the other, improving agent training on ALFWorld and WebShop.
-
LibEvoBench: Probing Temporal Knowledge Stratification in Code Generation Models
LibEvoBench benchmark shows LLMs are version-oblivious on evolving APIs, with documentation helping but version specification not.
-
PreAct: Computer-Using Agents that Get Faster on Repeated Tasks
PreAct compiles successful agent executions into verifiable state-machine programs for 8.5-13x faster replay on repeated tasks, with an independent evaluator check before storing each program.
-
MetaSyn: A Benchmark for LLM Agents on Meta-Analysis Articles from Nature Portfolio
MetaSyn is a stage-level benchmark of 442 meta-analyses showing LLM agents retrieve up to 90.9% of eligible studies but include at most 52.7% in their final reports.
-
Memory Beyond Recall: A Dual-Process Cognitive Memory System for Self-Evolving LLM Agents
DCPM reorganizes LLM agent memory into a cognitive hierarchy driven by a synchronous daytime belief writer and an asynchronous nighttime schema engine, reporting gains on cross-session inference benchmarks.
-
Re-Ranking Through an Attribution Lens for Citation Quality in Legal QA
Re-ranking retrieval candidates via a cross-encoder trained on continuous perturbation-based attribution scores improves citation faithfulness and gold-answer alignment in legal QA over semantic similarity.
-
Tool-Schema Compression Enables Agentic RAG Under Constrained Context Budgets
Tool schema compression by 44-50% enables agentic RAG at 8K context where uncompressed schemas fail, with +20.5 pp exact match lift across models and scaling to over 800 tools.
-
Spectral Retrieval: Multi-Scale Sinc Convolution over Token Embeddings for Localized Retrieval in LLM Multi-Agent Systems
Spectral Retrieval uses multi-scale sinc convolutions on token embeddings to interpolate between per-token MaxSim and mean-pooling, achieving large gains on synthetic and LIMIT-small benchmarks for localized retrieval.
-
IdioLink: Retrieving Meaning Beyond Words Across Idiomatic and Literal Expressions
IdioLink introduces a benchmark dataset and evaluation showing that strong embedding models struggle to retrieve equivalent meanings across idiomatic and literal forms, relying on shallow cues instead.
-
PluRule: A Benchmark for Moderating Pluralistic Communities on Social Media
PluRule is a new multimodal multilingual benchmark showing that state-of-the-art vision-language models perform only marginally better than a trivial baseline at detecting specific rule violations in pluralistic online communities.
-
Remember the Decision, Not the Description: A Rate-Distortion Framework for Agent Memory
Memory for long-horizon agents should preserve distinctions that affect decisions under a fixed budget, not descriptive features, yielding an exact forgetting boundary and a new online learner DeMem with regret guarantees.
-
MemFlow: Intent-Driven Memory Orchestration for Small Language Model Agents
MemFlow routes queries by intent to tiered memory operations, nearly doubling accuracy of a 1.7B SLM on long-horizon benchmarks compared to full-context baselines.
-
Enhancing Agent Safety Judgment: Controlled Benchmark Rewriting and Analogical Reasoning for Deceptive Out-of-Distribution Scenarios
ROME generates deceptive safety benchmarks that degrade LLM agent judgment performance, while ARISE uses analogical retrieval to improve safety decisions at inference time without retraining.
-
Perturbation Dose Responses in Recursive LLM Loops: Raw Switching, Stochastic Floors, and Persistent Escape under Append, Replace, and Dialog Updates
In 30-step recursive LLM loops, append-mode persistent escape from source basins reaches 50% near 400 tokens under full history but plateaus below 50% under tail-clip memory policy, while replace-mode switching largely reflects state reset.
-
OCR-Memory: Optical Context Retrieval for Long-Horizon Agent Memory
OCR-Memory encodes agent trajectories as images with visual anchors and retrieves verbatim text via locate-and-transcribe, yielding gains on long-horizon benchmarks under strict context limits.
-
Training Transformers as a Universal Computer
A transformer trained on random meaningless MicroPy programs generalizes to execute diverse human-written programs, providing empirical evidence it can act as a universal computer.
-
XGRAG: A Graph-Native Framework for Explaining KG-based Retrieval-Augmented Generation
XGRAG uses graph perturbations to quantify component contributions in GraphRAG and achieves 14.81% better explanation quality than text-based baselines on QA datasets, with correlations to graph centrality.
-
Similar Users-Augmented Interest Network
SUIN improves CTR prediction by augmenting target user sequences with similar users' behaviors via embedding-based retrieval, user-specific position encoding, and user-aware target attention.
-
A Systematic Survey of Security Threats and Defenses in LLM-Based AI Agents: A Layered Attack Surface Framework
A new 7x4 taxonomy organizes agentic AI security threats by architectural layer and persistence timescale, revealing under-explored upper layers and missing defenses after surveying 116 papers.
-
Dr.Sai: An agentic AI for real-world physics analysis at BESIII
Dr.Sai autonomously executed full physics analysis pipelines on real BESIII data to re-measure ten J/psi decay branching fractions, matching established benchmarks without any manual coding.
-
Learning When Not to Decide: A Framework for Overcoming Factual Presumptuousness in AI Adjudication
A new structured prompting method (SPEC) helps AI detect insufficient evidence in adjudication tasks and defer decisions appropriately, reaching 89% accuracy on a benchmark varying information completeness from Colorado unemployment insurance cases.
-
From Experience to Skill: Multi-Agent Generative Engine Optimization via Reusable Strategy Learning
MAGEO is a multi-agent system that distills validated editing patterns into reusable optimization skills for generative engines, outperforming heuristic baselines on visibility and fidelity via a new benchmark and evaluation protocol.
-
RAGognizer: Hallucination-Aware Fine-Tuning via Detection Head Integration
RAGognizer adds a detection head to LLMs for joint training on generation and token-level hallucination detection, yielding SOTA detection and fewer hallucinations in RAG while preserving output quality.
-
IG-Search: Step-Level Information Gain Rewards for Search-Augmented Reasoning
IG-Search computes step-level information gain rewards from policy probabilities to improve credit assignment in RL training for search-augmented QA, yielding 1.6-point gains over trajectory-level baselines on multi-hop tasks.
-
TimeSeriesExamAgent: Creating Time Series Reasoning Benchmarks at Scale
TimeSeriesExamAgent combines templates and LLM agents to generate scalable time series reasoning benchmarks, demonstrating that current LLMs have limited performance on both abstract and domain-specific tasks.
-
IoT-Brain: Grounding LLMs for Semantic-Spatial Sensor Scheduling
IoT-Brain uses a neuro-symbolic Spatial Trajectory Graph to ground LLMs for verifiable semantic-spatial sensor scheduling, achieving 37.6% higher task success with lower resource use on a campus-scale benchmark.
-
An Agentic Evaluation Architecture for Historical Bias Detection in Educational Textbooks
An agentic architecture with multimodal screening, a five-agent jury, meta-synthesis, and source attribution protocol detects biases in Romanian history textbooks more accurately than zero-shot baselines, achieving 83.3% acceptable excerpts and human preference in 64.8% of blind comparisons.
-
SkillGraph: Graph Foundation Priors for LLM Agent Tool Sequence Recommendation
SkillGraph builds a reusable execution-transition graph prior from LLM trajectories and applies it via hybrid retrieval plus learned reranking to raise tool-sequence quality on ToolBench and API-Bank benchmarks.
-
Springdrift: An Auditable Persistent Runtime for LLM Agents with Case-Based Memory, Normative Safety, and Ambient Self-Perception
Springdrift provides an auditable persistent runtime for long-lived LLM agents with case-based memory, normative safety gating, and ambient self-perception, shown in a 23-day single-instance deployment where the agent self-diagnosed bugs and maintained cross-channel context.
-
Do We Still Need GraphRAG? Benchmarking RAG and GraphRAG for Agentic Search Systems
Agentic search narrows the gap between dense RAG and GraphRAG but does not remove GraphRAG's advantage on complex multi-hop reasoning.
-
CounterRefine: Answer-Conditioned Counterevidence Retrieval for Inference-Time Knowledge Repair in Factual Question Answering
CounterRefine improves factual QA accuracy by up to 5.8 points on SimpleQA through answer-conditioned counterevidence retrieval and validated refinement with minimal output changes.
-
AtomicRAG: Atom-Entity Graphs for Retrieval-Augmented Generation
AtomicRAG replaces chunk-based and triple-based GraphRAG with atom-entity graphs that store facts as atomic units and use personalized PageRank plus relevance filtering to achieve higher retrieval accuracy and reasoning robustness on five benchmarks.
-
Compass vs Railway Tracks: Unpacking User Mental Models for Communicating Long-Horizon Work to Humans vs. AI
Users treat human delegation for long tasks as a flexible compass but AI delegation as rigid railway tracks due to perceived AI limitations in inference and judgment.
-
V-RoAst: Visual Road Assessment. Can VLM be a Road Safety Assessor Using the iRAP Standard?
V-RoAst applies zero-shot VLMs (Gemini-1.5-flash, GPT-4o-mini) to iRAP road safety attribute classification on a new ThaiRAP image dataset and compares them to CNN baselines, finding better generalization to unseen classes but weaker spatial reasoning.
-
AI Native Games: A Survey and Roadmap
AI-native games require runtime generative AI as a non-substitutable core-loop mechanism; a 53-game corpus clusters in language-forward narrative and epistemic designs.
-
Temporal Validity in Retrieval Memory: Eliminating Stale-Fact Errors for AI Agents over Evolving Knowledge
MemStrata maintains temporal validity in retrieval memory via deterministic (s,r,o) supersession, matching RAG on static recall while reaching 0.95-1.00 accuracy on evolving knowledge where RAG scores 0.20-0.47 and cutting stale-fact errors from 15-40% to near zero.
-
SAGE-Nav: Leveraging LLM Planning and Alignment Fusion for Hierarchical Scene Graph-Guided Navigation
SAGE-Nav decouples LLM global planning from reactive control via hierarchical scene graphs and alignment fusion, reporting SOTA results on i-THOR and RoboTHOR with improved efficiency and zero-shot generalization.
-
Novelty-Aware Agentic Retrieval: Comparing Research Contributions Through Structured Multi-Step Reasoning
The Novelty-Aware Research Agent layers query analysis, ReAct retrieval, ranking, schema-guided extraction, three-pass comparison, and answer generation on RAG to produce structured comparison artifacts that standard RAG cannot.
-
CodeTeam: An LLM-Powered Multi-Agent Framework for Repository-Level Code Generation
CodeTeam is an LLM multi-agent system that improves SketchBLEU by 4.1/2.9 points and achieves top test pass rates (34.6% PE, 42.3% SFT) on repository-level code generation benchmarks via role-specialized planning and implementation stages.
-
A Unified Framework for Context-Aware and Relation-Aware Graph Retrieval-Augmented Generation
HyGRAG is a hierarchical graph RAG framework that constructs LLM summaries over hybrid chunk-entity graphs, retrieves via context and relation awareness across levels, and enables dynamic updates, reporting a 9.7% average accuracy gain on multi-hop reasoning tasks.
-
Selection Integrity for LLM Graph Memory: An Accumulability Criterion for Information-Flow-Blind Retrieval
Provenance checks in graph memory are blind to structural attacks that reallocate top-k membership; authselect prevents this by enforcing selection on the authenticated subgraph only.
-
Substrate Asymmetry in User-Side Memory: A Diagnostic Framework
User memory in LLMs factors into three orthogonal axes where parametric adapters and retrieval show opposite strengths, with causal evidence from attention interventions and an alignment tax on RLHF models.
-
The Agentic Web Requires New Normative Infrastructure
The web's anti-bot regime should be replaced by a framework that presumptively lets user-authorized AI agents act for their principals, requires platforms to disclose access policies, and permits agent blocking only when proportionate to concrete harms.
-
MMClima: A Framework for Multimodal Climate Science Data and Evaluation
MMClima releases a 104k+ multimodal climate QA dataset built via automated extraction plus human validation, benchmarks multimodal LLMs, and releases a fine-tuned 70B textual model that outperforms baselines.