{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FVDTH55FUB5QJNOH6IMSOQXQNY","short_pith_number":"pith:FVDTH55F","schema_version":"1.0","canonical_sha256":"2d4733f7a5a07b04b5c7f2192742f06e3ab99a301de997a167e8f328bf88879c","source":{"kind":"arxiv","id":"2406.14972","version":1},"attestation_state":"computed","paper":{"title":"A Tale of Trust and Accuracy: Base vs. Instruct LLMs in RAG Systems","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Fabrizio Silvestri, Florin Cuconasu, Giovanni Trappolini, Nicola Tonellotto","submitted_at":"2024-06-21T08:31:02Z","abstract_excerpt":"Retrieval Augmented Generation (RAG) represents a significant advancement in artificial intelligence combining a retrieval phase with a generative phase, with the latter typically being powered by large language models (LLMs). The current common practices in RAG involve using \"instructed\" LLMs, which are fine-tuned with supervised training to enhance their ability to follow instructions and are aligned with human preferences using state-of-the-art techniques. Contrary to popular belief, our study demonstrates that base models outperform their instructed counterparts in RAG tasks by 20% on aver"},"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":"2406.14972","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-06-21T08:31:02Z","cross_cats_sorted":["cs.IR"],"title_canon_sha256":"b5066f4958a16a30c8e108b11e496214a9c52dae91d2a2f0c1e72cd8c4d61f95","abstract_canon_sha256":"4b56eca472f670c5510040249185542d00881279c99d63f5236acd53e52cffbb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:35:11.387254Z","signature_b64":"3lEF2K/XnQ4V9sSHW/KFOWkRdeLyG52vboP+vrzCb8dAsO48cF+vk+G4yseDjKf/4ov4WY8A3JXO7EJyX21IDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2d4733f7a5a07b04b5c7f2192742f06e3ab99a301de997a167e8f328bf88879c","last_reissued_at":"2026-07-05T08:35:11.386868Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:35:11.386868Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Tale of Trust and Accuracy: Base vs. Instruct LLMs in RAG Systems","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IR"],"primary_cat":"cs.CL","authors_text":"Fabrizio Silvestri, Florin Cuconasu, Giovanni Trappolini, Nicola Tonellotto","submitted_at":"2024-06-21T08:31:02Z","abstract_excerpt":"Retrieval Augmented Generation (RAG) represents a significant advancement in artificial intelligence combining a retrieval phase with a generative phase, with the latter typically being powered by large language models (LLMs). The current common practices in RAG involve using \"instructed\" LLMs, which are fine-tuned with supervised training to enhance their ability to follow instructions and are aligned with human preferences using state-of-the-art techniques. Contrary to popular belief, our study demonstrates that base models outperform their instructed counterparts in RAG tasks by 20% on aver"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.14972","kind":"arxiv","version":1},"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/2406.14972/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":"2406.14972","created_at":"2026-07-05T08:35:11.386925+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.14972v1","created_at":"2026-07-05T08:35:11.386925+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.14972","created_at":"2026-07-05T08:35:11.386925+00:00"},{"alias_kind":"pith_short_12","alias_value":"FVDTH55FUB5Q","created_at":"2026-07-05T08:35:11.386925+00:00"},{"alias_kind":"pith_short_16","alias_value":"FVDTH55FUB5QJNOH","created_at":"2026-07-05T08:35:11.386925+00:00"},{"alias_kind":"pith_short_8","alias_value":"FVDTH55F","created_at":"2026-07-05T08:35:11.386925+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2411.19463","citing_title":"Understanding the Fundamental Design Decisions of Retrieval-Augmented Generation Systems","ref_index":15,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY","json":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY.json","graph_json":"https://pith.science/api/pith-number/FVDTH55FUB5QJNOH6IMSOQXQNY/graph.json","events_json":"https://pith.science/api/pith-number/FVDTH55FUB5QJNOH6IMSOQXQNY/events.json","paper":"https://pith.science/paper/FVDTH55F"},"agent_actions":{"view_html":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY","download_json":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY.json","view_paper":"https://pith.science/paper/FVDTH55F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.14972&json=true","fetch_graph":"https://pith.science/api/pith-number/FVDTH55FUB5QJNOH6IMSOQXQNY/graph.json","fetch_events":"https://pith.science/api/pith-number/FVDTH55FUB5QJNOH6IMSOQXQNY/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY/action/storage_attestation","attest_author":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY/action/author_attestation","sign_citation":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY/action/citation_signature","submit_replication":"https://pith.science/pith/FVDTH55FUB5QJNOH6IMSOQXQNY/action/replication_record"}},"created_at":"2026-07-05T08:35:11.386925+00:00","updated_at":"2026-07-05T08:35:11.386925+00:00"}