{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5TRTD5FWYGKYHYXJHXALAV5S5F","short_pith_number":"pith:5TRTD5FW","schema_version":"1.0","canonical_sha256":"ece331f4b6c19583e2e93dc0b057b2e9567633317ed0b074b9b1ed96b42d11de","source":{"kind":"arxiv","id":"2408.15300","version":1},"attestation_state":"computed","paper":{"title":"GIFT-SW: Gaussian noise Injected Fine-Tuning of Salient Weights for LLMs","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aleksandr Zuev, Egor Shvetsov, Egor Venediktov, Evgeny Burnaev, Mariya Krylova, Maxim Zhelnin, Viktor Moskvoretskii","submitted_at":"2024-08-27T14:41:14Z","abstract_excerpt":"Parameter Efficient Fine-Tuning (PEFT) methods have gained popularity and democratized the usage of Large Language Models (LLMs). Recent studies have shown that a small subset of weights significantly impacts performance. Based on this observation, we introduce a novel PEFT method, called Gaussian noise Injected Fine Tuning of Salient Weights (GIFT-SW). Our method updates only salient columns, while injecting Gaussian noise into non-salient ones. To identify these columns, we developeda generalized sensitivity metric that extends and unifies metrics from previous studies. Experiments with LLaM"},"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":"2408.15300","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.LG","submitted_at":"2024-08-27T14:41:14Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"7b0e17ee86a78ac7907e85630feaad6c470e98a6cbf37be60c63e86defca18f7","abstract_canon_sha256":"e94b40fc293c90d66e3fbb2a8b09d42a140fa99f7fced36bd7b0959d67f9b01b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:00:09.990858Z","signature_b64":"VycM2QwCgxMG8YxbTP/kF2c0SNsSwVm/kR4f3xYIBi9D9Dp1saI3LGnm7YcTx9wd38l/u6qyoI/Ld2dfle2qDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ece331f4b6c19583e2e93dc0b057b2e9567633317ed0b074b9b1ed96b42d11de","last_reissued_at":"2026-07-05T09:00:09.990430Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:00:09.990430Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GIFT-SW: Gaussian noise Injected Fine-Tuning of Salient Weights for LLMs","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Aleksandr Zuev, Egor Shvetsov, Egor Venediktov, Evgeny Burnaev, Mariya Krylova, Maxim Zhelnin, Viktor Moskvoretskii","submitted_at":"2024-08-27T14:41:14Z","abstract_excerpt":"Parameter Efficient Fine-Tuning (PEFT) methods have gained popularity and democratized the usage of Large Language Models (LLMs). Recent studies have shown that a small subset of weights significantly impacts performance. Based on this observation, we introduce a novel PEFT method, called Gaussian noise Injected Fine Tuning of Salient Weights (GIFT-SW). Our method updates only salient columns, while injecting Gaussian noise into non-salient ones. To identify these columns, we developeda generalized sensitivity metric that extends and unifies metrics from previous studies. Experiments with LLaM"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2408.15300","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/2408.15300/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":"2408.15300","created_at":"2026-07-05T09:00:09.990482+00:00"},{"alias_kind":"arxiv_version","alias_value":"2408.15300v1","created_at":"2026-07-05T09:00:09.990482+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2408.15300","created_at":"2026-07-05T09:00:09.990482+00:00"},{"alias_kind":"pith_short_12","alias_value":"5TRTD5FWYGKY","created_at":"2026-07-05T09:00:09.990482+00:00"},{"alias_kind":"pith_short_16","alias_value":"5TRTD5FWYGKYHYXJ","created_at":"2026-07-05T09:00:09.990482+00:00"},{"alias_kind":"pith_short_8","alias_value":"5TRTD5FW","created_at":"2026-07-05T09:00:09.990482+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2507.03052","citing_title":"From 2:4 to 8:16 sparsity patterns in LLMs for Outliers and Weights with Variance Correction","ref_index":34,"is_internal_anchor":false},{"citing_arxiv_id":"2509.22166","citing_title":"Motivating Next-Gen Accelerators with Flexible (N:M) Activation Sparsity via Benchmarking Lightweight Post-Training Sparsification Approaches","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F","json":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F.json","graph_json":"https://pith.science/api/pith-number/5TRTD5FWYGKYHYXJHXALAV5S5F/graph.json","events_json":"https://pith.science/api/pith-number/5TRTD5FWYGKYHYXJHXALAV5S5F/events.json","paper":"https://pith.science/paper/5TRTD5FW"},"agent_actions":{"view_html":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F","download_json":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F.json","view_paper":"https://pith.science/paper/5TRTD5FW","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2408.15300&json=true","fetch_graph":"https://pith.science/api/pith-number/5TRTD5FWYGKYHYXJHXALAV5S5F/graph.json","fetch_events":"https://pith.science/api/pith-number/5TRTD5FWYGKYHYXJHXALAV5S5F/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F/action/storage_attestation","attest_author":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F/action/author_attestation","sign_citation":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F/action/citation_signature","submit_replication":"https://pith.science/pith/5TRTD5FWYGKYHYXJHXALAV5S5F/action/replication_record"}},"created_at":"2026-07-05T09:00:09.990482+00:00","updated_at":"2026-07-05T09:00:09.990482+00:00"}