{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:HNKWCCATJNNRJWNCZIJ63NS2CI","short_pith_number":"pith:HNKWCCAT","schema_version":"1.0","canonical_sha256":"3b556108134b5b14d9a2ca13edb65a1238f6b381dad0f5b071286552726306e5","source":{"kind":"arxiv","id":"2410.20008","version":1},"attestation_state":"computed","paper":{"title":"Layer by Layer: Uncovering Where Multi-Task Learning Happens in Instruction-Tuned Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Shay B. Cohen, Yftah Ziser, Zheng Zhao","submitted_at":"2024-10-25T23:38:28Z","abstract_excerpt":"Fine-tuning pre-trained large language models (LLMs) on a diverse array of tasks has become a common approach for building models that can solve various natural language processing (NLP) tasks. However, where and to what extent these models retain task-specific knowledge remains largely unexplored. This study investigates the task-specific information encoded in pre-trained LLMs and the effects of instruction tuning on their representations across a diverse set of over 60 NLP tasks. We use a set of matrix analysis tools to examine the differences between the way pre-trained and instruction-tun"},"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":"2410.20008","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2024-10-25T23:38:28Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"d4b32c888320acd8712b8062080848c9c1285e7fbe505958205148517197c605","abstract_canon_sha256":"347400bb7e0e1531bed171e746511547407510ef3665e2877490fc4544c35396"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:26:37.426369Z","signature_b64":"wYepxnFqMq/qylXiN781aRaLAkPq8yn5Q0grC/bdB3yq9/HEBeHk9XRs2kfGZknnjqxZsspfFBMnPwXvA4AODQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3b556108134b5b14d9a2ca13edb65a1238f6b381dad0f5b071286552726306e5","last_reissued_at":"2026-07-05T09:26:37.425846Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:26:37.425846Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Layer by Layer: Uncovering Where Multi-Task Learning Happens in Instruction-Tuned Large Language Models","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Shay B. Cohen, Yftah Ziser, Zheng Zhao","submitted_at":"2024-10-25T23:38:28Z","abstract_excerpt":"Fine-tuning pre-trained large language models (LLMs) on a diverse array of tasks has become a common approach for building models that can solve various natural language processing (NLP) tasks. However, where and to what extent these models retain task-specific knowledge remains largely unexplored. This study investigates the task-specific information encoded in pre-trained LLMs and the effects of instruction tuning on their representations across a diverse set of over 60 NLP tasks. We use a set of matrix analysis tools to examine the differences between the way pre-trained and instruction-tun"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.20008","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/2410.20008/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":"2410.20008","created_at":"2026-07-05T09:26:37.425910+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.20008v1","created_at":"2026-07-05T09:26:37.425910+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.20008","created_at":"2026-07-05T09:26:37.425910+00:00"},{"alias_kind":"pith_short_12","alias_value":"HNKWCCATJNNR","created_at":"2026-07-05T09:26:37.425910+00:00"},{"alias_kind":"pith_short_16","alias_value":"HNKWCCATJNNRJWNC","created_at":"2026-07-05T09:26:37.425910+00:00"},{"alias_kind":"pith_short_8","alias_value":"HNKWCCAT","created_at":"2026-07-05T09:26:37.425910+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2604.08939","citing_title":"Delve into the Applicability of Advanced Optimizers for Multi-Task Learning","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI","json":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI.json","graph_json":"https://pith.science/api/pith-number/HNKWCCATJNNRJWNCZIJ63NS2CI/graph.json","events_json":"https://pith.science/api/pith-number/HNKWCCATJNNRJWNCZIJ63NS2CI/events.json","paper":"https://pith.science/paper/HNKWCCAT"},"agent_actions":{"view_html":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI","download_json":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI.json","view_paper":"https://pith.science/paper/HNKWCCAT","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.20008&json=true","fetch_graph":"https://pith.science/api/pith-number/HNKWCCATJNNRJWNCZIJ63NS2CI/graph.json","fetch_events":"https://pith.science/api/pith-number/HNKWCCATJNNRJWNCZIJ63NS2CI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI/action/storage_attestation","attest_author":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI/action/author_attestation","sign_citation":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI/action/citation_signature","submit_replication":"https://pith.science/pith/HNKWCCATJNNRJWNCZIJ63NS2CI/action/replication_record"}},"created_at":"2026-07-05T09:26:37.425910+00:00","updated_at":"2026-07-05T09:26:37.425910+00:00"}