{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:GRDKFQOFOLU7GCE3VBHQRGDGS2","short_pith_number":"pith:GRDKFQOF","schema_version":"1.0","canonical_sha256":"3446a2c1c572e9f3089ba84f089866968af066b50d4bdad95d15e975fe331521","source":{"kind":"arxiv","id":"2501.04070","version":3},"attestation_state":"computed","paper":{"title":"More is not always better? Enhancing Many-Shot In-Context Learning with Differentiated and Reweighting Objectives","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ang Lv, Flood Sung, Jian Luan, Rui Yan, Shuo Shang, Wei Liu, Xiaoqing Zhang, Xiuying Chen, Yuhan Liu","submitted_at":"2025-01-07T14:57:08Z","abstract_excerpt":"Large language models (LLMs) excel at few-shot in-context learning (ICL) without requiring parameter updates. However, as ICL demonstrations increase from a few to many, performance tends to plateau and eventually decline. We identify two primary causes for this trend: the suboptimal negative log-likelihood (NLL) optimization objective and the incremental data noise. To address these issues, we introduce \\textit{DrICL}, a novel optimization method that enhances model performance through \\textit{Differentiated} and \\textit{Reweighting} objectives. Globally, DrICL utilizes differentiated learnin"},"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":"2501.04070","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2025-01-07T14:57:08Z","cross_cats_sorted":["cs.AI","cs.CL"],"title_canon_sha256":"49b8841bfe6b5045c69556f0db637f131865cc194acdf91a768d85164a73caa3","abstract_canon_sha256":"1578b6ec7f5d9fc9abb23d6863854b20f130a83bfc7cddd823bf678f3ff5a1f6"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:10:03.829341Z","signature_b64":"/yNVXqpocqqJ3SS3F6kw/FaxgM91oeBsFli/kB8bLb4qjtlQRwAe4TRKPC2oLYXhL5+qedRNgQd0BF03F0cCCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3446a2c1c572e9f3089ba84f089866968af066b50d4bdad95d15e975fe331521","last_reissued_at":"2026-07-05T11:10:03.828827Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:10:03.828827Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"More is not always better? Enhancing Many-Shot In-Context Learning with Differentiated and Reweighting Objectives","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI","cs.CL"],"primary_cat":"cs.LG","authors_text":"Ang Lv, Flood Sung, Jian Luan, Rui Yan, Shuo Shang, Wei Liu, Xiaoqing Zhang, Xiuying Chen, Yuhan Liu","submitted_at":"2025-01-07T14:57:08Z","abstract_excerpt":"Large language models (LLMs) excel at few-shot in-context learning (ICL) without requiring parameter updates. However, as ICL demonstrations increase from a few to many, performance tends to plateau and eventually decline. We identify two primary causes for this trend: the suboptimal negative log-likelihood (NLL) optimization objective and the incremental data noise. To address these issues, we introduce \\textit{DrICL}, a novel optimization method that enhances model performance through \\textit{Differentiated} and \\textit{Reweighting} objectives. Globally, DrICL utilizes differentiated learnin"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.04070","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/2501.04070/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":"2501.04070","created_at":"2026-07-05T11:10:03.828885+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.04070v3","created_at":"2026-07-05T11:10:03.828885+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.04070","created_at":"2026-07-05T11:10:03.828885+00:00"},{"alias_kind":"pith_short_12","alias_value":"GRDKFQOFOLU7","created_at":"2026-07-05T11:10:03.828885+00:00"},{"alias_kind":"pith_short_16","alias_value":"GRDKFQOFOLU7GCE3","created_at":"2026-07-05T11:10:03.828885+00:00"},{"alias_kind":"pith_short_8","alias_value":"GRDKFQOF","created_at":"2026-07-05T11:10:03.828885+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2504.02181","citing_title":"A Survey of Scaling in Large Language Model Reasoning","ref_index":253,"is_internal_anchor":false},{"citing_arxiv_id":"2507.20906","citing_title":"Soft Head Selection for Injecting ICL-Derived Task Embeddings","ref_index":25,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2","json":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2.json","graph_json":"https://pith.science/api/pith-number/GRDKFQOFOLU7GCE3VBHQRGDGS2/graph.json","events_json":"https://pith.science/api/pith-number/GRDKFQOFOLU7GCE3VBHQRGDGS2/events.json","paper":"https://pith.science/paper/GRDKFQOF"},"agent_actions":{"view_html":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2","download_json":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2.json","view_paper":"https://pith.science/paper/GRDKFQOF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.04070&json=true","fetch_graph":"https://pith.science/api/pith-number/GRDKFQOFOLU7GCE3VBHQRGDGS2/graph.json","fetch_events":"https://pith.science/api/pith-number/GRDKFQOFOLU7GCE3VBHQRGDGS2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2/action/storage_attestation","attest_author":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2/action/author_attestation","sign_citation":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2/action/citation_signature","submit_replication":"https://pith.science/pith/GRDKFQOFOLU7GCE3VBHQRGDGS2/action/replication_record"}},"created_at":"2026-07-05T11:10:03.828885+00:00","updated_at":"2026-07-05T11:10:03.828885+00:00"}