{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:K7B6KVOEPALYAOHO7WIIGLG5PJ","short_pith_number":"pith:K7B6KVOE","schema_version":"1.0","canonical_sha256":"57c3e555c478178038eefd90832cdd7a502b19f55832bf3405d6a3bc115b9949","source":{"kind":"arxiv","id":"2510.07118","version":3},"attestation_state":"computed","paper":{"title":"TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Deepak Ravikumar, Kaushik Roy, Manish Nagaraj, Sakshi Choudhary, Utkarsh Saxena","submitted_at":"2025-10-08T15:11:04Z","abstract_excerpt":"Instruction tuning is essential for aligning large language models (LLMs) to downstream tasks and commonly relies on large, diverse corpora. However, small, high-quality subsets, known as coresets, can deliver comparable or superior results, though curating them remains challenging. Existing methods often rely on coarse, sample-level signals like gradients, an approach that is computationally expensive and overlooks fine-grained features. To address this, we introduce TRIM (Token Relevance via Interpretable Multi-layer Attention), a forward-only, token-centric framework. Instead of using gradi"},"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":"2510.07118","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-10-08T15:11:04Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"63a862b66de70958eac89c3f3ae55b780eb71e2d9459f4c14f9927047a0a2dfd","abstract_canon_sha256":"1d3280112c8b7dac00db804db6a1d7801edc6bba1333f960289e9277fa566c85"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:01.076760Z","signature_b64":"EhPVVhhp1PqHTZ/XxW2EHrHvBjxyWuIRc9OX9qG8cBmsFvLU1rCWWSsLzfsaN35sr5JJd+scG2ams25eONecDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"57c3e555c478178038eefd90832cdd7a502b19f55832bf3405d6a3bc115b9949","last_reissued_at":"2026-05-17T23:39:01.076039Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:01.076039Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Deepak Ravikumar, Kaushik Roy, Manish Nagaraj, Sakshi Choudhary, Utkarsh Saxena","submitted_at":"2025-10-08T15:11:04Z","abstract_excerpt":"Instruction tuning is essential for aligning large language models (LLMs) to downstream tasks and commonly relies on large, diverse corpora. However, small, high-quality subsets, known as coresets, can deliver comparable or superior results, though curating them remains challenging. Existing methods often rely on coarse, sample-level signals like gradients, an approach that is computationally expensive and overlooks fine-grained features. To address this, we introduce TRIM (Token Relevance via Interpretable Multi-layer Attention), a forward-only, token-centric framework. Instead of using gradi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2510.07118","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":""},"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":"2510.07118","created_at":"2026-05-17T23:39:01.076170+00:00"},{"alias_kind":"arxiv_version","alias_value":"2510.07118v3","created_at":"2026-05-17T23:39:01.076170+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2510.07118","created_at":"2026-05-17T23:39:01.076170+00:00"},{"alias_kind":"pith_short_12","alias_value":"K7B6KVOEPALY","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"K7B6KVOEPALYAOHO","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"K7B6KVOE","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ","json":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ.json","graph_json":"https://pith.science/api/pith-number/K7B6KVOEPALYAOHO7WIIGLG5PJ/graph.json","events_json":"https://pith.science/api/pith-number/K7B6KVOEPALYAOHO7WIIGLG5PJ/events.json","paper":"https://pith.science/paper/K7B6KVOE"},"agent_actions":{"view_html":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ","download_json":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ.json","view_paper":"https://pith.science/paper/K7B6KVOE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2510.07118&json=true","fetch_graph":"https://pith.science/api/pith-number/K7B6KVOEPALYAOHO7WIIGLG5PJ/graph.json","fetch_events":"https://pith.science/api/pith-number/K7B6KVOEPALYAOHO7WIIGLG5PJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ/action/storage_attestation","attest_author":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ/action/author_attestation","sign_citation":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ/action/citation_signature","submit_replication":"https://pith.science/pith/K7B6KVOEPALYAOHO7WIIGLG5PJ/action/replication_record"}},"created_at":"2026-05-17T23:39:01.076170+00:00","updated_at":"2026-05-17T23:39:01.076170+00:00"}