{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2023:6T43IYZ4LFVEQECWYZ5RV6QZWK","short_pith_number":"pith:6T43IYZ4","canonical_record":{"source":{"id":"2302.06600","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-13T18:55:52Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ddbc611bb50fcfe3873619cdc922f383b1c5759d4cbb163d21cf966331ebee03","abstract_canon_sha256":"74c214428247f010a36c5a9e5539571ef5a4c876e3b2b090e22bb2e2f52980b4"},"schema_version":"1.0"},"canonical_sha256":"f4f9b4633c596a481056c67b1afa19b2a33da00ee121f8fe161c6fc0ce245f61","source":{"kind":"arxiv","id":"2302.06600","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2302.06600","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"arxiv_version","alias_value":"2302.06600v2","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.06600","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"pith_short_12","alias_value":"6T43IYZ4LFVE","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"pith_short_16","alias_value":"6T43IYZ4LFVEQECW","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"pith_short_8","alias_value":"6T43IYZ4","created_at":"2026-07-05T06:26:45Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2023:6T43IYZ4LFVEQECWYZ5RV6QZWK","target":"record","payload":{"canonical_record":{"source":{"id":"2302.06600","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-13T18:55:52Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"ddbc611bb50fcfe3873619cdc922f383b1c5759d4cbb163d21cf966331ebee03","abstract_canon_sha256":"74c214428247f010a36c5a9e5539571ef5a4c876e3b2b090e22bb2e2f52980b4"},"schema_version":"1.0"},"canonical_sha256":"f4f9b4633c596a481056c67b1afa19b2a33da00ee121f8fe161c6fc0ce245f61","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:26:45.816043Z","signature_b64":"OEktPe7XwUVRoWHZdQOb2KNhSjAsFxV/7aeJX1eSvoDMvqyZwyM8AS+mPdbROpxO5ClI8xq/b8iGgR1ap+rWDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f4f9b4633c596a481056c67b1afa19b2a33da00ee121f8fe161c6fc0ce245f61","last_reissued_at":"2026-07-05T06:26:45.815545Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:26:45.815545Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2302.06600","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T06:26:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"izWU2hM42Dxa3Lq1VVvkK/++YSCReH+AN8YDUF6y6B+V7GTs6/aphrp08CpWUCXIJBLTD355cAglZylxHOQBBw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T07:40:00.539999Z"},"content_sha256":"55d9f831380b41f1ba24f7681926aaf1d38af6f0ae78fa2b07ee1b1ca93bdf11","schema_version":"1.0","event_id":"sha256:55d9f831380b41f1ba24f7681926aaf1d38af6f0ae78fa2b07ee1b1ca93bdf11"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2023:6T43IYZ4LFVEQECWYZ5RV6QZWK","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Task-Specific Skill Localization in Fine-tuned Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Abhishek Panigrahi, Haoyu Zhao, Nikunj Saunshi, Sanjeev Arora","submitted_at":"2023-02-13T18:55:52Z","abstract_excerpt":"Pre-trained language models can be fine-tuned to solve diverse NLP tasks, including in few-shot settings. Thus fine-tuning allows the model to quickly pick up task-specific ``skills,'' but there has been limited study of where these newly-learnt skills reside inside the massive model. This paper introduces the term skill localization for this problem and proposes a solution. Given the downstream task and a model fine-tuned on that task, a simple optimization is used to identify a very small subset of parameters ($\\sim0.01$% of model parameters) responsible for ($>95$%) of the model's performan"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.06600","kind":"arxiv","version":2},"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/2302.06600/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T06:26:45Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"uacS+goPqwLtJFm89zOvUgo5+RJWO4DE6GpQy28WFhXFFpoAK71cXTgnxbmR/KOj0Go7sf/oTSPc/VbSmirvAg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-17T07:40:00.540645Z"},"content_sha256":"7fd1134492cdec8aa2d5757bc03fcc9d2df8177689ff5a02782f259b9b5af191","schema_version":"1.0","event_id":"sha256:7fd1134492cdec8aa2d5757bc03fcc9d2df8177689ff5a02782f259b9b5af191"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/6T43IYZ4LFVEQECWYZ5RV6QZWK/bundle.json","state_url":"https://pith.science/pith/6T43IYZ4LFVEQECWYZ5RV6QZWK/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/6T43IYZ4LFVEQECWYZ5RV6QZWK/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-17T07:40:00Z","links":{"resolver":"https://pith.science/pith/6T43IYZ4LFVEQECWYZ5RV6QZWK","bundle":"https://pith.science/pith/6T43IYZ4LFVEQECWYZ5RV6QZWK/bundle.json","state":"https://pith.science/pith/6T43IYZ4LFVEQECWYZ5RV6QZWK/state.json","well_known_bundle":"https://pith.science/.well-known/pith/6T43IYZ4LFVEQECWYZ5RV6QZWK/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:6T43IYZ4LFVEQECWYZ5RV6QZWK","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"74c214428247f010a36c5a9e5539571ef5a4c876e3b2b090e22bb2e2f52980b4","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-13T18:55:52Z","title_canon_sha256":"ddbc611bb50fcfe3873619cdc922f383b1c5759d4cbb163d21cf966331ebee03"},"schema_version":"1.0","source":{"id":"2302.06600","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2302.06600","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"arxiv_version","alias_value":"2302.06600v2","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.06600","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"pith_short_12","alias_value":"6T43IYZ4LFVE","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"pith_short_16","alias_value":"6T43IYZ4LFVEQECW","created_at":"2026-07-05T06:26:45Z"},{"alias_kind":"pith_short_8","alias_value":"6T43IYZ4","created_at":"2026-07-05T06:26:45Z"}],"graph_snapshots":[{"event_id":"sha256:7fd1134492cdec8aa2d5757bc03fcc9d2df8177689ff5a02782f259b9b5af191","target":"graph","created_at":"2026-07-05T06:26:45Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2302.06600/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Pre-trained language models can be fine-tuned to solve diverse NLP tasks, including in few-shot settings. Thus fine-tuning allows the model to quickly pick up task-specific ``skills,'' but there has been limited study of where these newly-learnt skills reside inside the massive model. This paper introduces the term skill localization for this problem and proposes a solution. Given the downstream task and a model fine-tuned on that task, a simple optimization is used to identify a very small subset of parameters ($\\sim0.01$% of model parameters) responsible for ($>95$%) of the model's performan","authors_text":"Abhishek Panigrahi, Haoyu Zhao, Nikunj Saunshi, Sanjeev Arora","cross_cats":["cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-13T18:55:52Z","title":"Task-Specific Skill Localization in Fine-tuned Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2302.06600","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:55d9f831380b41f1ba24f7681926aaf1d38af6f0ae78fa2b07ee1b1ca93bdf11","target":"record","created_at":"2026-07-05T06:26:45Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"74c214428247f010a36c5a9e5539571ef5a4c876e3b2b090e22bb2e2f52980b4","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-02-13T18:55:52Z","title_canon_sha256":"ddbc611bb50fcfe3873619cdc922f383b1c5759d4cbb163d21cf966331ebee03"},"schema_version":"1.0","source":{"id":"2302.06600","kind":"arxiv","version":2}},"canonical_sha256":"f4f9b4633c596a481056c67b1afa19b2a33da00ee121f8fe161c6fc0ce245f61","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f4f9b4633c596a481056c67b1afa19b2a33da00ee121f8fe161c6fc0ce245f61","first_computed_at":"2026-07-05T06:26:45.815545Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T06:26:45.815545Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"OEktPe7XwUVRoWHZdQOb2KNhSjAsFxV/7aeJX1eSvoDMvqyZwyM8AS+mPdbROpxO5ClI8xq/b8iGgR1ap+rWDQ==","signature_status":"signed_v1","signed_at":"2026-07-05T06:26:45.816043Z","signed_message":"canonical_sha256_bytes"},"source_id":"2302.06600","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:55d9f831380b41f1ba24f7681926aaf1d38af6f0ae78fa2b07ee1b1ca93bdf11","sha256:7fd1134492cdec8aa2d5757bc03fcc9d2df8177689ff5a02782f259b9b5af191"],"state_sha256":"9fbb973cfd7a3381a420ac82df1cf0aa9f4e637c27fa6d10ac2e77475e8a4f20"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"pJ0FEIRWm/8Vsy9zfPKKlZvzzgRdpsk1rCG1WdYMzbXa2PUhFaQek984t2R/y6Y1E8HIowYcjiWaNX6NB4wzBA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-17T07:40:00.561816Z","bundle_sha256":"7606696b5b49107cc155414509fb6ecf060c6b2abdb53c50bc967498adf4191e"}}