{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:NEQVXOG2XQKNKQNLC45PPHLLV6","short_pith_number":"pith:NEQVXOG2","schema_version":"1.0","canonical_sha256":"69215bb8dabc14d541ab173af79d6baf96c5bc00dafac7601b9e8dd2cfc5a47b","source":{"kind":"arxiv","id":"2407.04528","version":4},"attestation_state":"computed","paper":{"title":"GPT vs RETRO: Exploring the Intersection of Retrieval and Parameter-Efficient Fine-Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Aleksander Ficek, Jiaqi Zeng, Oleksii Kuchaiev","submitted_at":"2024-07-05T14:16:47Z","abstract_excerpt":"Parameter-Efficient Fine-Tuning (PEFT) and Retrieval-Augmented Generation (RAG) have become popular methods for adapting large language models while minimizing compute requirements. In this paper, we apply PEFT methods (P-tuning, Adapters, and LoRA) to a modified Retrieval-Enhanced Transformer (RETRO) and a baseline GPT model across several sizes, ranging from 823 million to 48 billion parameters. We show that RETRO models outperform GPT models in zero-shot settings due to their unique pre-training process but GPT models have higher performance potential with PEFT. Additionally, our study indi"},"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":"2407.04528","kind":"arxiv","version":4},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-07-05T14:16:47Z","cross_cats_sorted":["cs.AI","cs.IR","cs.LG"],"title_canon_sha256":"786df197bcecb40b0b6867e7a84f5d5d4dae732cc28041a779746ff39a275174","abstract_canon_sha256":"a799e4c6a912d38ec1132fe8d9db6b01e23adde46671659aefa00aea9b80c6cd"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:25:45.062131Z","signature_b64":"9YSyJaPgi11PCTyqyI/lGqvtJxujwbo/vtDRJG45Dbp33ZBuBmC9PUAkBrstlHgHqict3+MtTIF0OY0iiQpFDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"69215bb8dabc14d541ab173af79d6baf96c5bc00dafac7601b9e8dd2cfc5a47b","last_reissued_at":"2026-07-05T09:25:45.061640Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:25:45.061640Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GPT vs RETRO: Exploring the Intersection of Retrieval and Parameter-Efficient Fine-Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.IR","cs.LG"],"primary_cat":"cs.CL","authors_text":"Aleksander Ficek, Jiaqi Zeng, Oleksii Kuchaiev","submitted_at":"2024-07-05T14:16:47Z","abstract_excerpt":"Parameter-Efficient Fine-Tuning (PEFT) and Retrieval-Augmented Generation (RAG) have become popular methods for adapting large language models while minimizing compute requirements. In this paper, we apply PEFT methods (P-tuning, Adapters, and LoRA) to a modified Retrieval-Enhanced Transformer (RETRO) and a baseline GPT model across several sizes, ranging from 823 million to 48 billion parameters. We show that RETRO models outperform GPT models in zero-shot settings due to their unique pre-training process but GPT models have higher performance potential with PEFT. Additionally, our study indi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2407.04528","kind":"arxiv","version":4},"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/2407.04528/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":"2407.04528","created_at":"2026-07-05T09:25:45.061698+00:00"},{"alias_kind":"arxiv_version","alias_value":"2407.04528v4","created_at":"2026-07-05T09:25:45.061698+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2407.04528","created_at":"2026-07-05T09:25:45.061698+00:00"},{"alias_kind":"pith_short_12","alias_value":"NEQVXOG2XQKN","created_at":"2026-07-05T09:25:45.061698+00:00"},{"alias_kind":"pith_short_16","alias_value":"NEQVXOG2XQKNKQNL","created_at":"2026-07-05T09:25:45.061698+00:00"},{"alias_kind":"pith_short_8","alias_value":"NEQVXOG2","created_at":"2026-07-05T09:25:45.061698+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2505.22921","citing_title":"Structured Memory Mechanisms for Stable Context Representation in Large Language Models","ref_index":20,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6","json":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6.json","graph_json":"https://pith.science/api/pith-number/NEQVXOG2XQKNKQNLC45PPHLLV6/graph.json","events_json":"https://pith.science/api/pith-number/NEQVXOG2XQKNKQNLC45PPHLLV6/events.json","paper":"https://pith.science/paper/NEQVXOG2"},"agent_actions":{"view_html":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6","download_json":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6.json","view_paper":"https://pith.science/paper/NEQVXOG2","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2407.04528&json=true","fetch_graph":"https://pith.science/api/pith-number/NEQVXOG2XQKNKQNLC45PPHLLV6/graph.json","fetch_events":"https://pith.science/api/pith-number/NEQVXOG2XQKNKQNLC45PPHLLV6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6/action/storage_attestation","attest_author":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6/action/author_attestation","sign_citation":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6/action/citation_signature","submit_replication":"https://pith.science/pith/NEQVXOG2XQKNKQNLC45PPHLLV6/action/replication_record"}},"created_at":"2026-07-05T09:25:45.061698+00:00","updated_at":"2026-07-05T09:25:45.061698+00:00"}