{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:RWT45WHEDQ2T77UZTST5ALRBRS","short_pith_number":"pith:RWT45WHE","schema_version":"1.0","canonical_sha256":"8da7ced8e41c353ffe999ca7d02e218cb8181a55c59ecfcb1bb850111953d964","source":{"kind":"arxiv","id":"2605.03229","version":2},"attestation_state":"computed","paper":{"title":"Sparse Memory Finetuning as a Low-Forgetting Alternative to LoRA and Full Finetuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Sparse Memory Finetuning updates only the most heavily read rows in added key-value layers to gain task performance while preserving general capabilities better than LoRA or full finetuning.","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Anirudh Kanchi, Garv Shah, Prakhar Gupta, Satyam Goyal","submitted_at":"2026-05-04T23:46:40Z","abstract_excerpt":"Adapting a pretrained language model to a new task often hurts the general capabilities it already had, a problem known as catastrophic forgetting. Sparse Memory Finetuning (SMF) tries to avoid this by adding key-value memory layers to the model and, on each training step, updating only the small set of memory rows that the current batch reads most heavily. We re-implement SMF on Qwen-2.5-0.5B-Instruct and compare it with LoRA and full finetuning on MedMCQA, a 4-choice medical exam task, using WikiText perplexity and TriviaQA accuracy as forgetting probes. SMF improves MedMCQA by 2.5 percentag"},"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":true},"canonical_record":{"source":{"id":"2605.03229","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-05-04T23:46:40Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"a9199ad6290c362e71d7ebdeb0f59de454c47aeaa7e76e814a1cd0e7cca22e0a","abstract_canon_sha256":"f5f1f66b2cd960f80551f3f7b38a468e1641e83de9f26e9393a906e31f0a5c58"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T02:08:43.619774Z","signature_b64":"E7ClZE8pNBaS461kdtsvkZ1FCIIkRjPpnQqPaZaQVLpIBebpGsUwFz/JJuND7C2J0QyQ+IaNqm80UI1njlwIDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8da7ced8e41c353ffe999ca7d02e218cb8181a55c59ecfcb1bb850111953d964","last_reissued_at":"2026-06-09T02:08:43.618522Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T02:08:43.618522Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Sparse Memory Finetuning as a Low-Forgetting Alternative to LoRA and Full Finetuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Sparse Memory Finetuning updates only the most heavily read rows in added key-value layers to gain task performance while preserving general capabilities better than LoRA or full finetuning.","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Anirudh Kanchi, Garv Shah, Prakhar Gupta, Satyam Goyal","submitted_at":"2026-05-04T23:46:40Z","abstract_excerpt":"Adapting a pretrained language model to a new task often hurts the general capabilities it already had, a problem known as catastrophic forgetting. Sparse Memory Finetuning (SMF) tries to avoid this by adding key-value memory layers to the model and, on each training step, updating only the small set of memory rows that the current batch reads most heavily. We re-implement SMF on Qwen-2.5-0.5B-Instruct and compare it with LoRA and full finetuning on MedMCQA, a 4-choice medical exam task, using WikiText perplexity and TriviaQA accuracy as forgetting probes. SMF improves MedMCQA by 2.5 percentag"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"SMF improves MedMCQA by 2.5 percentage points while keeping both forgetting probes within roughly 1 point of the base model, whereas LoRA and full finetuning achieve larger gains but with clear drift on both.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That selectively updating only the most heavily read memory rows during training is sufficient to acquire new task knowledge without unintended interference in unrelated general capabilities.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"SMF improves MedMCQA accuracy by 2.5 points while keeping WikiText perplexity and TriviaQA accuracy within 1 point of the base model, outperforming LoRA and full finetuning on forgetting metrics.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Sparse Memory Finetuning updates only the most heavily read rows in added key-value layers to gain task performance while preserving general capabilities better than LoRA or full finetuning.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f936b72255b9524f1282e03c8b02357a9ae1149b2ebc7e71d4028d621d89a9c2"},"source":{"id":"2605.03229","kind":"arxiv","version":2},"verdict":{"id":"919f7bab-02fa-49fa-9ecc-267ca47dbda6","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-08T17:47:39.561915Z","strongest_claim":"SMF improves MedMCQA by 2.5 percentage points while keeping both forgetting probes within roughly 1 point of the base model, whereas LoRA and full finetuning achieve larger gains but with clear drift on both.","one_line_summary":"SMF improves MedMCQA accuracy by 2.5 points while keeping WikiText perplexity and TriviaQA accuracy within 1 point of the base model, outperforming LoRA and full finetuning on forgetting metrics.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That selectively updating only the most heavily read memory rows during training is sufficient to acquire new task knowledge without unintended interference in unrelated general capabilities.","pith_extraction_headline":"Sparse Memory Finetuning updates only the most heavily read rows in added key-value layers to gain task performance while preserving general capabilities better than LoRA or full finetuning."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.03229/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T14:35:19.247524Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-20T01:31:21.794774Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T15:34:04.280909Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"6dc37e4ce03a15165ed85bcb415ba044c98eea186f08319dd73876c48ed8e5a9"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"58dce1ef058b469a5bb57b12b9a65cce0e644b5409789449ba5ae62c061a9f55"},"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":"2605.03229","created_at":"2026-06-09T02:08:43.618654+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.03229v2","created_at":"2026-06-09T02:08:43.618654+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.03229","created_at":"2026-06-09T02:08:43.618654+00:00"},{"alias_kind":"pith_short_12","alias_value":"RWT45WHEDQ2T","created_at":"2026-06-09T02:08:43.618654+00:00"},{"alias_kind":"pith_short_16","alias_value":"RWT45WHEDQ2T77UZ","created_at":"2026-06-09T02:08:43.618654+00:00"},{"alias_kind":"pith_short_8","alias_value":"RWT45WHE","created_at":"2026-06-09T02:08:43.618654+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":2,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS","json":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS.json","graph_json":"https://pith.science/api/pith-number/RWT45WHEDQ2T77UZTST5ALRBRS/graph.json","events_json":"https://pith.science/api/pith-number/RWT45WHEDQ2T77UZTST5ALRBRS/events.json","paper":"https://pith.science/paper/RWT45WHE"},"agent_actions":{"view_html":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS","download_json":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS.json","view_paper":"https://pith.science/paper/RWT45WHE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.03229&json=true","fetch_graph":"https://pith.science/api/pith-number/RWT45WHEDQ2T77UZTST5ALRBRS/graph.json","fetch_events":"https://pith.science/api/pith-number/RWT45WHEDQ2T77UZTST5ALRBRS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS/action/storage_attestation","attest_author":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS/action/author_attestation","sign_citation":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS/action/citation_signature","submit_replication":"https://pith.science/pith/RWT45WHEDQ2T77UZTST5ALRBRS/action/replication_record"}},"created_at":"2026-06-09T02:08:43.618654+00:00","updated_at":"2026-06-09T02:08:43.618654+00:00"}