{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:64SV5BSOZBP3Z432G3DIVXP2FH","short_pith_number":"pith:64SV5BSO","canonical_record":{"source":{"id":"2512.01766","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-01T15:08:43Z","cross_cats_sorted":[],"title_canon_sha256":"4c1784338706bb0534f91fa5c9a711ef284fd1913be679bb249fca138a7e1b02","abstract_canon_sha256":"ac75b42556910ed9a5a2a132fcb3231d6691f4b187e09d64cb8ee5f3262283f3"},"schema_version":"1.0"},"canonical_sha256":"f7255e864ec85fbcf37a36c68addfa29c1200a39eb8de91d656f6b98fc79e1c4","source":{"kind":"arxiv","id":"2512.01766","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.01766","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"arxiv_version","alias_value":"2512.01766v2","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.01766","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"pith_short_12","alias_value":"64SV5BSOZBP3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"64SV5BSOZBP3Z432","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"64SV5BSO","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:64SV5BSOZBP3Z432G3DIVXP2FH","target":"record","payload":{"canonical_record":{"source":{"id":"2512.01766","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-01T15:08:43Z","cross_cats_sorted":[],"title_canon_sha256":"4c1784338706bb0534f91fa5c9a711ef284fd1913be679bb249fca138a7e1b02","abstract_canon_sha256":"ac75b42556910ed9a5a2a132fcb3231d6691f4b187e09d64cb8ee5f3262283f3"},"schema_version":"1.0"},"canonical_sha256":"f7255e864ec85fbcf37a36c68addfa29c1200a39eb8de91d656f6b98fc79e1c4","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:16.975084Z","signature_b64":"JRqxxpv+Oe4cu0kCHXPsZbvpsvJoKEUBfxQ9Bst2MvykL9NWVemK0haUpjp8lg3KMOMkVmm05Ci9LQO0eUauDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f7255e864ec85fbcf37a36c68addfa29c1200a39eb8de91d656f6b98fc79e1c4","last_reissued_at":"2026-05-17T23:39:16.974390Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:16.974390Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2512.01766","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":"2026-05-17T23:39:16Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"h5MtTyisIGb5MrxOk18D+t5YlVvxBitvCX9cy9kb/l7SLxskfseNHoR71J2dDtw+gWknOBUPIFYxF8aDFGyTCQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-08T13:40:39.137599Z"},"content_sha256":"96749aae67a6ea72a147258a2150114d2d05e6beb570552226079f2fd095c5f4","schema_version":"1.0","event_id":"sha256:96749aae67a6ea72a147258a2150114d2d05e6beb570552226079f2fd095c5f4"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:64SV5BSOZBP3Z432G3DIVXP2FH","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"On the Unreasonable Effectiveness of Last-layer Retraining","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Last-layer retraining succeeds mainly because the held-out set has better group balance than the full training data.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"John C. Hill, Tyler LaBonte, Vidya Muthukumar, Xinchen Zhang","submitted_at":"2025-12-01T15:08:43Z","abstract_excerpt":"Last-layer retraining (LLR) methods -- wherein the last layer of a neural network is reinitialized and retrained on a held-out set following ERM training -- have garnered interest as an efficient approach to rectify dependence on spurious correlations and improve performance on minority groups. Surprisingly, LLR has been found to improve worst-group accuracy even when the held-out set is an imbalanced subset of the training set. We initially hypothesize that this ``unreasonable effectiveness'' of LLR is explained by its ability to mitigate neural collapse through the held-out set, resulting in"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"We present strong evidence for an alternative hypothesis: that the success of LLR is primarily due to better group balance in the held-out set.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the empirical tests on neural collapse are conclusive enough to rule it out as a contributing factor, and that group balance differences are the dominant cause rather than other unmeasured effects.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Last-layer retraining succeeds primarily because the held-out set has better group balance, supported by experiments that do not back the neural collapse hypothesis.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Last-layer retraining succeeds mainly because the held-out set has better group balance than the full training data.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0b7b3eb0347cea5ea0b1bf0d3f0057551bbffbe3ec6c61c48e1da7a07544aaf7"},"source":{"id":"2512.01766","kind":"arxiv","version":2},"verdict":{"id":"f57ac31a-9834-42de-a042-3529860fa6c4","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-17T02:39:21.315834Z","strongest_claim":"We present strong evidence for an alternative hypothesis: that the success of LLR is primarily due to better group balance in the held-out set.","one_line_summary":"Last-layer retraining succeeds primarily because the held-out set has better group balance, supported by experiments that do not back the neural collapse hypothesis.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the empirical tests on neural collapse are conclusive enough to rule it out as a contributing factor, and that group balance differences are the dominant cause rather than other unmeasured effects.","pith_extraction_headline":"Last-layer retraining succeeds mainly because the held-out set has better group balance than the full training data."},"references":{"count":4,"sample":[{"doi":"10.1073/pnas.2103091118.url:http://dx.doi.org/10.1073/","year":1998,"title":"Boosting the margin: A new explanation for the effectiveness of voting methods","work_id":"18928c48-673b-4b24-8e73-8e2efe177a6d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2018,"title":"A broad-coverage challenge corpus for sentence understanding through inference","work_id":"ab1ad3fa-51ee-4b7d-ba96-8f51078051f8","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1992,"title":"Note that Waterbirds is the only dataset that has a distribution shift and MultiNLI is the only dataset which is class-balanceda priori","work_id":"2f29390d-9bd5-42f4-9409-90fa8f21e077","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"These pretrained models are used as the initialization for ERM finetuning under the cross-entropy loss","work_id":"279fb77b-4749-4b2f-93d8-9068c2720b07","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":4,"snapshot_sha256":"cce3bac278234ceef620b01009cf6196a9d54eef76e558c96f07bb8e267df06d","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"02ffadf4474dddc8cb5b9e114b7b7cc721cf2646dcbd60582f5b320aab2fcbb6"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"f57ac31a-9834-42de-a042-3529860fa6c4"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-17T23:39:16Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"xsO0+nMQ2fGcBBUCIT6o8MeN7CnTa8Nr1oVeDZScg4WJ/DZ+a8b4XnHDu+pnS5f03FGW6w6T1XDmL+p9DQInCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-08T13:40:39.138082Z"},"content_sha256":"8d1072c9efd3612bc3e91de7d90363e27c4de8bbb406c140fbe2a97f8ee4e574","schema_version":"1.0","event_id":"sha256:8d1072c9efd3612bc3e91de7d90363e27c4de8bbb406c140fbe2a97f8ee4e574"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/64SV5BSOZBP3Z432G3DIVXP2FH/bundle.json","state_url":"https://pith.science/pith/64SV5BSOZBP3Z432G3DIVXP2FH/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/64SV5BSOZBP3Z432G3DIVXP2FH/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":"2026-06-08T13:40:39Z","links":{"resolver":"https://pith.science/pith/64SV5BSOZBP3Z432G3DIVXP2FH","bundle":"https://pith.science/pith/64SV5BSOZBP3Z432G3DIVXP2FH/bundle.json","state":"https://pith.science/pith/64SV5BSOZBP3Z432G3DIVXP2FH/state.json","well_known_bundle":"https://pith.science/.well-known/pith/64SV5BSOZBP3Z432G3DIVXP2FH/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:64SV5BSOZBP3Z432G3DIVXP2FH","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":"ac75b42556910ed9a5a2a132fcb3231d6691f4b187e09d64cb8ee5f3262283f3","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-01T15:08:43Z","title_canon_sha256":"4c1784338706bb0534f91fa5c9a711ef284fd1913be679bb249fca138a7e1b02"},"schema_version":"1.0","source":{"id":"2512.01766","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.01766","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"arxiv_version","alias_value":"2512.01766v2","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.01766","created_at":"2026-05-17T23:39:16Z"},{"alias_kind":"pith_short_12","alias_value":"64SV5BSOZBP3","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"64SV5BSOZBP3Z432","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"64SV5BSO","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:8d1072c9efd3612bc3e91de7d90363e27c4de8bbb406c140fbe2a97f8ee4e574","target":"graph","created_at":"2026-05-17T23:39:16Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"We present strong evidence for an alternative hypothesis: that the success of LLR is primarily due to better group balance in the held-out set."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That the empirical tests on neural collapse are conclusive enough to rule it out as a contributing factor, and that group balance differences are the dominant cause rather than other unmeasured effects."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Last-layer retraining succeeds primarily because the held-out set has better group balance, supported by experiments that do not back the neural collapse hypothesis."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Last-layer retraining succeeds mainly because the held-out set has better group balance than the full training data."}],"snapshot_sha256":"0b7b3eb0347cea5ea0b1bf0d3f0057551bbffbe3ec6c61c48e1da7a07544aaf7"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"02ffadf4474dddc8cb5b9e114b7b7cc721cf2646dcbd60582f5b320aab2fcbb6"},"paper":{"abstract_excerpt":"Last-layer retraining (LLR) methods -- wherein the last layer of a neural network is reinitialized and retrained on a held-out set following ERM training -- have garnered interest as an efficient approach to rectify dependence on spurious correlations and improve performance on minority groups. Surprisingly, LLR has been found to improve worst-group accuracy even when the held-out set is an imbalanced subset of the training set. We initially hypothesize that this ``unreasonable effectiveness'' of LLR is explained by its ability to mitigate neural collapse through the held-out set, resulting in","authors_text":"John C. Hill, Tyler LaBonte, Vidya Muthukumar, Xinchen Zhang","cross_cats":[],"headline":"Last-layer retraining succeeds mainly because the held-out set has better group balance than the full training data.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-01T15:08:43Z","title":"On the Unreasonable Effectiveness of Last-layer Retraining"},"references":{"count":4,"internal_anchors":0,"resolved_work":4,"sample":[{"cited_arxiv_id":"","doi":"10.1073/pnas.2103091118.url:http://dx.doi.org/10.1073/","is_internal_anchor":false,"ref_index":1,"title":"Boosting the margin: A new explanation for the effectiveness of voting methods","work_id":"18928c48-673b-4b24-8e73-8e2efe177a6d","year":1998},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"A broad-coverage challenge corpus for sentence understanding through inference","work_id":"ab1ad3fa-51ee-4b7d-ba96-8f51078051f8","year":2018},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Note that Waterbirds is the only dataset that has a distribution shift and MultiNLI is the only dataset which is class-balanceda priori","work_id":"2f29390d-9bd5-42f4-9409-90fa8f21e077","year":1992},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"These pretrained models are used as the initialization for ERM finetuning under the cross-entropy loss","work_id":"279fb77b-4749-4b2f-93d8-9068c2720b07","year":2022}],"snapshot_sha256":"cce3bac278234ceef620b01009cf6196a9d54eef76e558c96f07bb8e267df06d"},"source":{"id":"2512.01766","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-17T02:39:21.315834Z","id":"f57ac31a-9834-42de-a042-3529860fa6c4","model_set":{"reader":"grok-4.3"},"one_line_summary":"Last-layer retraining succeeds primarily because the held-out set has better group balance, supported by experiments that do not back the neural collapse hypothesis.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Last-layer retraining succeeds mainly because the held-out set has better group balance than the full training data.","strongest_claim":"We present strong evidence for an alternative hypothesis: that the success of LLR is primarily due to better group balance in the held-out set.","weakest_assumption":"That the empirical tests on neural collapse are conclusive enough to rule it out as a contributing factor, and that group balance differences are the dominant cause rather than other unmeasured effects."}},"verdict_id":"f57ac31a-9834-42de-a042-3529860fa6c4"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:96749aae67a6ea72a147258a2150114d2d05e6beb570552226079f2fd095c5f4","target":"record","created_at":"2026-05-17T23:39:16Z","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":"ac75b42556910ed9a5a2a132fcb3231d6691f4b187e09d64cb8ee5f3262283f3","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2025-12-01T15:08:43Z","title_canon_sha256":"4c1784338706bb0534f91fa5c9a711ef284fd1913be679bb249fca138a7e1b02"},"schema_version":"1.0","source":{"id":"2512.01766","kind":"arxiv","version":2}},"canonical_sha256":"f7255e864ec85fbcf37a36c68addfa29c1200a39eb8de91d656f6b98fc79e1c4","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"f7255e864ec85fbcf37a36c68addfa29c1200a39eb8de91d656f6b98fc79e1c4","first_computed_at":"2026-05-17T23:39:16.974390Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:16.974390Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"JRqxxpv+Oe4cu0kCHXPsZbvpsvJoKEUBfxQ9Bst2MvykL9NWVemK0haUpjp8lg3KMOMkVmm05Ci9LQO0eUauDA==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:16.975084Z","signed_message":"canonical_sha256_bytes"},"source_id":"2512.01766","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:96749aae67a6ea72a147258a2150114d2d05e6beb570552226079f2fd095c5f4","sha256:8d1072c9efd3612bc3e91de7d90363e27c4de8bbb406c140fbe2a97f8ee4e574"],"state_sha256":"5296d5ecdae733589ed24bd88d901fc70616243ec78bec901d0d8ba798f921ed"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"SCSkcU220Rkl55tg7ohHJmS+FwI7ebGzpix13I84S5z6c2LKmb5gZ4RN66BizSbOhdNJXDyW44JjRKcGbneuBw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-08T13:40:39.140400Z","bundle_sha256":"6fd62a9995189b06e3abdf4525ce686e606e3c57a401aee03b9b7f1b3c98f157"}}