{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:TY53PNBRXAMUJZJBEW2VW57JPT","short_pith_number":"pith:TY53PNBR","schema_version":"1.0","canonical_sha256":"9e3bb7b431b81944e52125b55b77e97cf00393ddc8e6350ca84e71b2687bbd10","source":{"kind":"arxiv","id":"2310.07713","version":3},"attestation_state":"computed","paper":{"title":"InstructRetro: Instruction Tuning post Retrieval-Augmented Pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bo Li, Boxin Wang, Bryan Catanzaro, Lawrence McAfee, Mohammad Shoeybi, Peng Xu, Wei Ping","submitted_at":"2023-10-11T17:59:05Z","abstract_excerpt":"Pretraining auto-regressive large language models~(LLMs) with retrieval demonstrates better perplexity and factual accuracy by leveraging external databases. However, the size of existing pretrained retrieval-augmented LLM is still limited (e.g., Retro has 7.5B parameters), which limits the effectiveness of instruction tuning and zero-shot generalization. In this work, we introduce Retro 48B, the largest LLM pretrained with retrieval. Specifically, we continue to pretrain a 43B GPT model on additional 100 billion tokens using the Retro augmentation method by retrieving from 1.2 trillion tokens"},"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":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2310.07713","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-10-11T17:59:05Z","cross_cats_sorted":["cs.AI","cs.IR","cs.LG"],"title_canon_sha256":"2204bcfd766a1c4c5c304d0c52c8e6d0eacf9b6efbb3fa9430618cf3285c0cdb","abstract_canon_sha256":"54b053ff2339f1e2681aea5fde3019105a89f205d8f69fdb3c469b119142ec34"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:24:22.725540Z","signature_b64":"/85S9mo0Dsr6A/B+z2idiMrUWB7iHutoa1LRR+W3S+IbHxarkl0AHgLGz86Tc0bvoX03hgRiWYgJMmRsk+ILDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9e3bb7b431b81944e52125b55b77e97cf00393ddc8e6350ca84e71b2687bbd10","last_reissued_at":"2026-07-05T08:24:22.725025Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:24:22.725025Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"InstructRetro: Instruction Tuning post Retrieval-Augmented Pretraining","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Bo Li, Boxin Wang, Bryan Catanzaro, Lawrence McAfee, Mohammad Shoeybi, Peng Xu, Wei Ping","submitted_at":"2023-10-11T17:59:05Z","abstract_excerpt":"Pretraining auto-regressive large language models~(LLMs) with retrieval demonstrates better perplexity and factual accuracy by leveraging external databases. However, the size of existing pretrained retrieval-augmented LLM is still limited (e.g., Retro has 7.5B parameters), which limits the effectiveness of instruction tuning and zero-shot generalization. In this work, we introduce Retro 48B, the largest LLM pretrained with retrieval. Specifically, we continue to pretrain a 43B GPT model on additional 100 billion tokens using the Retro augmentation method by retrieving from 1.2 trillion tokens"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2310.07713","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2310.07713/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"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":"2310.07713","created_at":"2026-07-05T08:24:22.725087+00:00"},{"alias_kind":"arxiv_version","alias_value":"2310.07713v3","created_at":"2026-07-05T08:24:22.725087+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2310.07713","created_at":"2026-07-05T08:24:22.725087+00:00"},{"alias_kind":"pith_short_12","alias_value":"TY53PNBRXAMU","created_at":"2026-07-05T08:24:22.725087+00:00"},{"alias_kind":"pith_short_16","alias_value":"TY53PNBRXAMUJZJB","created_at":"2026-07-05T08:24:22.725087+00:00"},{"alias_kind":"pith_short_8","alias_value":"TY53PNBR","created_at":"2026-07-05T08:24:22.725087+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":7,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2312.10997","citing_title":"Retrieval-Augmented Generation for Large Language Models: A Survey","ref_index":45,"is_internal_anchor":false},{"citing_arxiv_id":"2407.13193","citing_title":"Retrieval-Augmented Generation for Natural Language Processing: A Survey","ref_index":167,"is_internal_anchor":false},{"citing_arxiv_id":"2506.20670","citing_title":"MMSearch-R1: Incentivizing LMMs to Search","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2405.17428","citing_title":"NV-Embed: Improved Techniques for Training LLMs as Generalist Embedding Models","ref_index":123,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12813","citing_title":"REALISTA: Realistic Latent Adversarial Attacks that Elicit LLM Hallucinations","ref_index":120,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05806","citing_title":"Retrieval from Within: An Intrinsic Capability of Attention-Based Models","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05806","citing_title":"Retrieval from Within: An Intrinsic Capability of Attention-Based Models","ref_index":36,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT","json":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT.json","graph_json":"https://pith.science/api/pith-number/TY53PNBRXAMUJZJBEW2VW57JPT/graph.json","events_json":"https://pith.science/api/pith-number/TY53PNBRXAMUJZJBEW2VW57JPT/events.json","paper":"https://pith.science/paper/TY53PNBR"},"agent_actions":{"view_html":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT","download_json":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT.json","view_paper":"https://pith.science/paper/TY53PNBR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2310.07713&json=true","fetch_graph":"https://pith.science/api/pith-number/TY53PNBRXAMUJZJBEW2VW57JPT/graph.json","fetch_events":"https://pith.science/api/pith-number/TY53PNBRXAMUJZJBEW2VW57JPT/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT/action/storage_attestation","attest_author":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT/action/author_attestation","sign_citation":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT/action/citation_signature","submit_replication":"https://pith.science/pith/TY53PNBRXAMUJZJBEW2VW57JPT/action/replication_record"}},"created_at":"2026-07-05T08:24:22.725087+00:00","updated_at":"2026-07-05T08:24:22.725087+00:00"}