{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:S2TJTMPPYHW32D5ESW4VJ672LC","short_pith_number":"pith:S2TJTMPP","schema_version":"1.0","canonical_sha256":"96a699b1efc1edbd0fa495b954fbfa58bc4baefed86c2e351038dcb7ef6897de","source":{"kind":"arxiv","id":"2310.11511","version":1},"attestation_state":"computed","paper":{"title":"Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Self-RAG trains a single language model to adaptively retrieve passages on demand and critique its own outputs using special reflection tokens.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Akari Asai, Avirup Sil, Hannaneh Hajishirzi, Yizhong Wang, Zeqiu Wu","submitted_at":"2023-10-17T18:18:32Z","abstract_excerpt":"Despite their remarkable capabilities, large language models (LLMs) often produce responses containing factual inaccuracies due to their sole reliance on the parametric knowledge they encapsulate. Retrieval-Augmented Generation (RAG), an ad hoc approach that augments LMs with retrieval of relevant knowledge, decreases such issues. However, indiscriminately retrieving and incorporating a fixed number of retrieved passages, regardless of whether retrieval is necessary, or passages are relevant, diminishes LM versatility or can lead to unhelpful response generation. We introduce a new framework c"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":true,"formal_links_present":true},"canonical_record":{"source":{"id":"2310.11511","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-17T18:18:32Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"4a2dd2ad87d2a6160674ba66eff6276562daa1a531df578df1f5f1e675a0628c","abstract_canon_sha256":"561e44b2432b4f4e957118a069dde7fae8fc9bfbf8aac408de3d4dfd1beed17d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:02:08.375436Z","signature_b64":"KN1yHuMoUH4WIBNiSRsWEhAsg02IKbg+HmNY/wJzUMHqg6NdEswt6LtpHJnG/aJe4TnFwlIWiR5dFzSmLWU7Aw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"96a699b1efc1edbd0fa495b954fbfa58bc4baefed86c2e351038dcb7ef6897de","last_reissued_at":"2026-07-05T07:02:08.374924Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:02:08.374924Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Self-RAG trains a single language model to adaptively retrieve passages on demand and critique its own outputs using special reflection tokens.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Akari Asai, Avirup Sil, Hannaneh Hajishirzi, Yizhong Wang, Zeqiu Wu","submitted_at":"2023-10-17T18:18:32Z","abstract_excerpt":"Despite their remarkable capabilities, large language models (LLMs) often produce responses containing factual inaccuracies due to their sole reliance on the parametric knowledge they encapsulate. Retrieval-Augmented Generation (RAG), an ad hoc approach that augments LMs with retrieval of relevant knowledge, decreases such issues. However, indiscriminately retrieving and incorporating a fixed number of retrieved passages, regardless of whether retrieval is necessary, or passages are relevant, diminishes LM versatility or can lead to unhelpful response generation. We introduce a new framework c"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Self-RAG (7B and 13B parameters) significantly outperforms state-of-the-art LLMs and retrieval-augmented models on a diverse set of tasks. Specifically, Self-RAG outperforms ChatGPT and retrieval-augmented Llama2-chat on Open-domain QA, reasoning and fact verification tasks, and it shows significant gains in improving factuality and citation accuracy for long-form generations relative to these models.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That an arbitrary LM can be trained to generate and act on reflection tokens in a way that improves rather than degrades performance across tasks without introducing new failure modes from the added tokens.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Self-RAG trains LLMs to adaptively retrieve passages on demand and self-critique using reflection tokens, outperforming ChatGPT and retrieval-augmented Llama2 on QA, reasoning, and fact verification.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Self-RAG trains a single language model to adaptively retrieve passages on demand and critique its own outputs using special reflection tokens.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"23d566505dfd6411625c27004b8d998868e212164cf9d9736afc9c8f84a533f2"},"source":{"id":"2310.11511","kind":"arxiv","version":1},"verdict":{"id":"baa194fe-ec6e-45a6-bbd4-8745d6e57266","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-12T14:09:42.760870Z","strongest_claim":"Self-RAG (7B and 13B parameters) significantly outperforms state-of-the-art LLMs and retrieval-augmented models on a diverse set of tasks. Specifically, Self-RAG outperforms ChatGPT and retrieval-augmented Llama2-chat on Open-domain QA, reasoning and fact verification tasks, and it shows significant gains in improving factuality and citation accuracy for long-form generations relative to these models.","one_line_summary":"Self-RAG trains LLMs to adaptively retrieve passages on demand and self-critique using reflection tokens, outperforming ChatGPT and retrieval-augmented Llama2 on QA, reasoning, and fact verification.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That an arbitrary LM can be trained to generate and act on reflection tokens in a way that improves rather than degrades performance across tasks without introducing new failure modes from the added tokens.","pith_extraction_headline":"Self-RAG trains a single language model to adaptively retrieve passages on demand and critique its own outputs using special reflection tokens."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2310.11511/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":154,"sample":[{"doi":"","year":2020,"title":"Learning to retrieve reasoning paths over wikipedia graph for question answering","work_id":"2f4e6df6-c1e2-4f5e-80b4-e86b8e29dadd","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Retrieval-based language models and applications","work_id":"8539632b-8089-43ed-b215-005ead5faa6e","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Task-aware retrieval with instructions","work_id":"31b1a479-fc74-4ee5-8290-a68ea7acc584","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Flashattention: Fast and memory-efficient exact attention with io-awareness","work_id":"0a762bb5-5795-45ee-a82d-71bfb1e1f9a4","ref_index":7,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Chain-of-Verification Reduces Hallucination in Large Language Models","work_id":"e9e9983c-a72a-4864-9005-cd09bbd7876c","ref_index":8,"cited_arxiv_id":"2309.11495","is_internal_anchor":true}],"resolved_work":154,"snapshot_sha256":"57d13d5f94e8cf59559f8f5e742c799af4ccd15be3917592ec7e1650e834e232","internal_anchors":23},"formal_canon":{"evidence_count":3,"snapshot_sha256":"45cd1e37665bd876ce638c287082c7a51af5ea5aa951b2e3a37e679baf017c19"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2310.11511","created_at":"2026-07-05T07:02:08.374996+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.11511v1","created_at":"2026-07-05T07:02:08.374996+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.11511","created_at":"2026-07-05T07:02:08.374996+00:00"},{"alias_kind":"pith_short_12","alias_value":"S2TJTMPPYHW3","created_at":"2026-07-05T07:02:08.374996+00:00"},{"alias_kind":"pith_short_16","alias_value":"S2TJTMPPYHW32D5E","created_at":"2026-07-05T07:02:08.374996+00:00"},{"alias_kind":"pith_short_8","alias_value":"S2TJTMPP","created_at":"2026-07-05T07:02:08.374996+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":81,"internal_anchor_count":81,"sample":[{"citing_arxiv_id":"2607.08028","citing_title":"From Prompts to Contracts: Harness Engineering for Auditable Enterprise LLM Agents","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2607.08700","citing_title":"Do You Need a Frontier Model as a Citation Verifier? Benchmarking Rubric LLMs for Deep-Research Source Attribution","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2607.07676","citing_title":"SkillCenter: A Large-Scale Source-Grounded Skill Library for Autonomous AI Agents","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2606.11199","citing_title":"NightFeats @ MMU-RAGent NeurIPS 2025: A Context-Optimized Multi-Agent RAG System for the Text-to-Text Track","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2606.26511","citing_title":"Temporal Validity in Retrieval Memory: Eliminating Stale-Fact Errors for AI Agents over Evolving Knowledge","ref_index":11,"is_internal_anchor":true},{"citing_arxiv_id":"2606.22681","citing_title":"Only Ask What You Don't Know: Grounded Delta Planning for Efficient Multi-step RAG","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2606.22610","citing_title":"PaperClaw: Harnessing Agents for Autonomous Research and Human-in-the-Loop Refinement","ref_index":53,"is_internal_anchor":true},{"citing_arxiv_id":"2606.22151","citing_title":"Novelty-Aware Agentic Retrieval: Comparing Research Contributions Through Structured Multi-Step Reasoning","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2606.21917","citing_title":"Pre-Generation Hallucination Detection in Large Language Models via Soft-Target Attention Probing","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2606.20997","citing_title":"BioInsight: Multi-Agent Orchestration for Interactive Biomedical Knowledge Discovery","ref_index":22,"is_internal_anchor":true},{"citing_arxiv_id":"2606.17721","citing_title":"Understanding and Debugging Failures in N-Gram-Based Generative Retrieval","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2606.17328","citing_title":"MemTrace: Probing What Final Accuracy Misses in Long-Term Memory","ref_index":45,"is_internal_anchor":true},{"citing_arxiv_id":"2606.13814","citing_title":"TASR: Training-Free Adaptive Stopping for Iterative Retrieval","ref_index":1,"is_internal_anchor":true},{"citing_arxiv_id":"2607.01935","citing_title":"A-TMA: Decoupling State-Aware Memory Failures in Long-Term Agent Memory","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"2607.01935","citing_title":"A-TMA: Decoupling State-Aware Memory Failures in Long-Term Agent Memory","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2606.12587","citing_title":"Strategic Decision Support for AI Agents","ref_index":5,"is_internal_anchor":true},{"citing_arxiv_id":"2606.11712","citing_title":"Substrate Asymmetry in User-Side Memory: A Diagnostic Framework","ref_index":39,"is_internal_anchor":true},{"citing_arxiv_id":"2606.08151","citing_title":"Decision-Aware Memory Cards: Counterfactual-Inspired Context Selection and Compression for Tool-Using LLM Agents","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2607.00038","citing_title":"Stop Hand-Holding Your Coding Agent: Engineering the Loops that Replace Step-by-Step Prompting","ref_index":3,"is_internal_anchor":true},{"citing_arxiv_id":"2606.07924","citing_title":"Decoupling Semantics and Logic: A Training-Free Coarse-to-Fine Pipeline for Video Retrieval-Augmented Generation","ref_index":20,"is_internal_anchor":true},{"citing_arxiv_id":"2606.01065","citing_title":"Leyline: KV Cache Directives for Agentic Inference","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00432","citing_title":"Grounded Decoding: Retrieval-Anchored Probability Fusion for Faithful RAG","ref_index":30,"is_internal_anchor":true},{"citing_arxiv_id":"2605.25162","citing_title":"STREAM: A Data-Centric Framework for Mining High-Value Task-Oriented Dialogues from Streaming Media","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2605.27452","citing_title":"Fine-Tuning Vision-Language Models for Understanding Current Damage and Scoring Priority with Quality Guard Agent","ref_index":46,"is_internal_anchor":true},{"citing_arxiv_id":"2605.26165","citing_title":"Tool-Schema Compression Enables Agentic RAG Under Constrained Context Budgets","ref_index":4,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":3,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC","json":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC.json","graph_json":"https://pith.science/api/pith-number/S2TJTMPPYHW32D5ESW4VJ672LC/graph.json","events_json":"https://pith.science/api/pith-number/S2TJTMPPYHW32D5ESW4VJ672LC/events.json","paper":"https://pith.science/paper/S2TJTMPP"},"agent_actions":{"view_html":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC","download_json":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC.json","view_paper":"https://pith.science/paper/S2TJTMPP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.11511&json=true","fetch_graph":"https://pith.science/api/pith-number/S2TJTMPPYHW32D5ESW4VJ672LC/graph.json","fetch_events":"https://pith.science/api/pith-number/S2TJTMPPYHW32D5ESW4VJ672LC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC/action/storage_attestation","attest_author":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC/action/author_attestation","sign_citation":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC/action/citation_signature","submit_replication":"https://pith.science/pith/S2TJTMPPYHW32D5ESW4VJ672LC/action/replication_record"}},"created_at":"2026-07-05T07:02:08.374996+00:00","updated_at":"2026-07-05T07:02:08.374996+00:00"}