{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:DPX3HFM4JMI5HWJMOK7AKPZYLS","short_pith_number":"pith:DPX3HFM4","schema_version":"1.0","canonical_sha256":"1befb3959c4b11d3d92c72be053f385c9b5f85a911301f0f6f47f1af8aab9a1a","source":{"kind":"arxiv","id":"2507.09185","version":1},"attestation_state":"computed","paper":{"title":"Detecting and Pruning Prominent but Detrimental Neurons in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Ameen Ali, Ivan Titov, Lior Wolf, Shahar Katz","submitted_at":"2025-07-12T08:10:10Z","abstract_excerpt":"Large language models (LLMs) often develop learned mechanisms specialized to specific datasets, such as reliance on domain-specific correlations, which yield high-confidence predictions without generalizable reasoning. While beneficial in one setting, these dataset-specific mechanisms typically degrade performance when models encounter novel tasks or distributions. In this work, we introduce a fine-tuning approach designed to enhance generalization by identifying and pruning neurons associated with dataset-specific mechanisms in transformer-based LLMs. Our method employs Integrated Gradients t"},"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":"2507.09185","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-07-12T08:10:10Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"69ebff8b1a22508d987940ab813977c853ec8d465c525036f38f4057916b82c6","abstract_canon_sha256":"2427a985227f1c4cba038f844d7cffdf821754fc81edb50734aa364803bad7ce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:36:05.164499Z","signature_b64":"QVocVMg+uixOCXacVi/uta/urV8r+/+EAePu78FriXvx4Pa1aaYMkUCuWqNOMGX6yEC1kA6DTf7j8GfyQ17VDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"1befb3959c4b11d3d92c72be053f385c9b5f85a911301f0f6f47f1af8aab9a1a","last_reissued_at":"2026-07-05T11:36:05.164050Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:36:05.164050Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Detecting and Pruning Prominent but Detrimental Neurons in Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Ameen Ali, Ivan Titov, Lior Wolf, Shahar Katz","submitted_at":"2025-07-12T08:10:10Z","abstract_excerpt":"Large language models (LLMs) often develop learned mechanisms specialized to specific datasets, such as reliance on domain-specific correlations, which yield high-confidence predictions without generalizable reasoning. While beneficial in one setting, these dataset-specific mechanisms typically degrade performance when models encounter novel tasks or distributions. In this work, we introduce a fine-tuning approach designed to enhance generalization by identifying and pruning neurons associated with dataset-specific mechanisms in transformer-based LLMs. Our method employs Integrated Gradients t"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.09185","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/2507.09185/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":"2507.09185","created_at":"2026-07-05T11:36:05.164106+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.09185v1","created_at":"2026-07-05T11:36:05.164106+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.09185","created_at":"2026-07-05T11:36:05.164106+00:00"},{"alias_kind":"pith_short_12","alias_value":"DPX3HFM4JMI5","created_at":"2026-07-05T11:36:05.164106+00:00"},{"alias_kind":"pith_short_16","alias_value":"DPX3HFM4JMI5HWJM","created_at":"2026-07-05T11:36:05.164106+00:00"},{"alias_kind":"pith_short_8","alias_value":"DPX3HFM4","created_at":"2026-07-05T11:36:05.164106+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.18453","citing_title":"LLM Parameters for Math Across Languages: Shared or Separate?","ref_index":27,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS","json":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS.json","graph_json":"https://pith.science/api/pith-number/DPX3HFM4JMI5HWJMOK7AKPZYLS/graph.json","events_json":"https://pith.science/api/pith-number/DPX3HFM4JMI5HWJMOK7AKPZYLS/events.json","paper":"https://pith.science/paper/DPX3HFM4"},"agent_actions":{"view_html":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS","download_json":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS.json","view_paper":"https://pith.science/paper/DPX3HFM4","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.09185&json=true","fetch_graph":"https://pith.science/api/pith-number/DPX3HFM4JMI5HWJMOK7AKPZYLS/graph.json","fetch_events":"https://pith.science/api/pith-number/DPX3HFM4JMI5HWJMOK7AKPZYLS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS/action/storage_attestation","attest_author":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS/action/author_attestation","sign_citation":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS/action/citation_signature","submit_replication":"https://pith.science/pith/DPX3HFM4JMI5HWJMOK7AKPZYLS/action/replication_record"}},"created_at":"2026-07-05T11:36:05.164106+00:00","updated_at":"2026-07-05T11:36:05.164106+00:00"}