{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:BEDAAMS6U35QYFN4RWWFGL7GH4","short_pith_number":"pith:BEDAAMS6","schema_version":"1.0","canonical_sha256":"090600325ea6fb0c15bc8dac532fe63f14375f0d580600c3b14b1f90d9297962","source":{"kind":"arxiv","id":"2305.15076","version":2},"attestation_state":"computed","paper":{"title":"Meta-Learning Online Adaptation of Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chelsea Finn, Christopher D. Manning, Eric Mitchell, Nathan Hu","submitted_at":"2023-05-24T11:56:20Z","abstract_excerpt":"Large language models encode impressively broad world knowledge in their parameters. However, the knowledge in static language models falls out of date, limiting the model's effective \"shelf life.\" While online fine-tuning can reduce this degradation, we find that naively fine-tuning on a stream of documents leads to a low level of information uptake. We hypothesize that online fine-tuning does not sufficiently attend to important information. That is, the gradient signal from important tokens representing factual information is drowned out by the gradient from inherently noisy tokens, suggest"},"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":"2305.15076","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2023-05-24T11:56:20Z","cross_cats_sorted":[],"title_canon_sha256":"a83b91224e709ea93d19badf68b41ab3536c3cc348fe6159f20d057a6e03a850","abstract_canon_sha256":"0f198da5f940cbb4e565ae1eeff966eace79bd1f119ee9ce1f8e484fa48e0a99"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T07:03:26.015359Z","signature_b64":"nw6Mtn9kmHFz0J3G22yGE6630rS+3/xm0JrPKIswUjHp0XcbhfMm3YUi+tiwBNeIRHBwe77BJBdRV78xyhHuCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"090600325ea6fb0c15bc8dac532fe63f14375f0d580600c3b14b1f90d9297962","last_reissued_at":"2026-07-05T07:03:26.014900Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T07:03:26.014900Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Meta-Learning Online Adaptation of Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Chelsea Finn, Christopher D. Manning, Eric Mitchell, Nathan Hu","submitted_at":"2023-05-24T11:56:20Z","abstract_excerpt":"Large language models encode impressively broad world knowledge in their parameters. However, the knowledge in static language models falls out of date, limiting the model's effective \"shelf life.\" While online fine-tuning can reduce this degradation, we find that naively fine-tuning on a stream of documents leads to a low level of information uptake. We hypothesize that online fine-tuning does not sufficiently attend to important information. That is, the gradient signal from important tokens representing factual information is drowned out by the gradient from inherently noisy tokens, suggest"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.15076","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/2305.15076/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":"2305.15076","created_at":"2026-07-05T07:03:26.014950+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.15076v2","created_at":"2026-07-05T07:03:26.014950+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.15076","created_at":"2026-07-05T07:03:26.014950+00:00"},{"alias_kind":"pith_short_12","alias_value":"BEDAAMS6U35Q","created_at":"2026-07-05T07:03:26.014950+00:00"},{"alias_kind":"pith_short_16","alias_value":"BEDAAMS6U35QYFN4","created_at":"2026-07-05T07:03:26.014950+00:00"},{"alias_kind":"pith_short_8","alias_value":"BEDAAMS6","created_at":"2026-07-05T07:03:26.014950+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4","json":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4.json","graph_json":"https://pith.science/api/pith-number/BEDAAMS6U35QYFN4RWWFGL7GH4/graph.json","events_json":"https://pith.science/api/pith-number/BEDAAMS6U35QYFN4RWWFGL7GH4/events.json","paper":"https://pith.science/paper/BEDAAMS6"},"agent_actions":{"view_html":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4","download_json":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4.json","view_paper":"https://pith.science/paper/BEDAAMS6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.15076&json=true","fetch_graph":"https://pith.science/api/pith-number/BEDAAMS6U35QYFN4RWWFGL7GH4/graph.json","fetch_events":"https://pith.science/api/pith-number/BEDAAMS6U35QYFN4RWWFGL7GH4/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4/action/timestamp_anchor","attest_storage":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4/action/storage_attestation","attest_author":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4/action/author_attestation","sign_citation":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4/action/citation_signature","submit_replication":"https://pith.science/pith/BEDAAMS6U35QYFN4RWWFGL7GH4/action/replication_record"}},"created_at":"2026-07-05T07:03:26.014950+00:00","updated_at":"2026-07-05T07:03:26.014950+00:00"}