{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:L7NYTNSTZJXPIILKGOVVKEXZ54","short_pith_number":"pith:L7NYTNST","schema_version":"1.0","canonical_sha256":"5fdb89b653ca6ef4216a33ab5512f9ef29b3b23ece549439da00d82227a108e7","source":{"kind":"arxiv","id":"2605.06152","version":3},"attestation_state":"computed","paper":{"title":"Grokking or Glitching? How Low-Precision Drives Slingshot Loss Spikes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Floating-point precision limits trigger slingshot loss spikes by creating numerical feature inflation in neural network training.","cross_cats":["cs.CL","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Jianjun Cao, Liu Hanqing, Yuanze Li, Zijian Zhou","submitted_at":"2026-05-07T12:45:21Z","abstract_excerpt":"Deep neural networks exhibit periodic loss spikes during unregularized long-term training, a phenomenon known as the \"Slingshot Mechanism.\" Existing work usually attributes this to intrinsic optimization dynamics, but its triggering mechanism remains unclear. This paper proves that this phenomenon is a result of floating-point arithmetic precision limits. As training enters a high-confidence stage, the difference between the correct-class logit and the other logits may exceed the absorption-error threshold. Then during backpropagation, the gradient of the correct class is rounded exactly to ze"},"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":"2605.06152","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-05-07T12:45:21Z","cross_cats_sorted":["cs.CL","math.OC","stat.ML"],"title_canon_sha256":"b286a045e76c4eba4a7313944c8d26b736e3eb958e213f3b7b49d9e25c003486","abstract_canon_sha256":"30d360ecf2d6eab91e508e8d66faefea61f8ca43dab8def0f9c1ed9e16f7486a"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:05:56.308281Z","signature_b64":"YNphSdfvrdNPvNgVizQj3z+bV3pAVjwAOnKQUtl8pzW/Bwf39ZL/UVOzKKMUKbi4HGeSTylPWd0bpgZJTaAjAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5fdb89b653ca6ef4216a33ab5512f9ef29b3b23ece549439da00d82227a108e7","last_reissued_at":"2026-05-27T01:05:56.307519Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:05:56.307519Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Grokking or Glitching? How Low-Precision Drives Slingshot Loss Spikes","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Floating-point precision limits trigger slingshot loss spikes by creating numerical feature inflation in neural network training.","cross_cats":["cs.CL","math.OC","stat.ML"],"primary_cat":"cs.LG","authors_text":"Jianjun Cao, Liu Hanqing, Yuanze Li, Zijian Zhou","submitted_at":"2026-05-07T12:45:21Z","abstract_excerpt":"Deep neural networks exhibit periodic loss spikes during unregularized long-term training, a phenomenon known as the \"Slingshot Mechanism.\" Existing work usually attributes this to intrinsic optimization dynamics, but its triggering mechanism remains unclear. This paper proves that this phenomenon is a result of floating-point arithmetic precision limits. As training enters a high-confidence stage, the difference between the correct-class logit and the other logits may exceed the absorption-error threshold. Then during backpropagation, the gradient of the correct class is rounded exactly to ze"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"This paper proves that this phenomenon is a result of floating-point arithmetic precision limits. ... We prove that this drift forms a positive feedback loop with the feature, causing the global classifier mean and the global feature mean to grow exponentially.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The assumption that, once the logit difference exceeds the absorption-error threshold, the gradient of the correct class is rounded exactly to zero during backpropagation while incorrect-class gradients remain nonzero, and that this imbalance necessarily creates an exponential positive feedback loop with the features.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Slingshot loss spikes arise from floating-point precision limits that round correct-class gradients to zero, breaking zero-sum constraints and driving exponential parameter growth through numerical feature inflation.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Floating-point precision limits trigger slingshot loss spikes by creating numerical feature inflation in neural network training.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"4403ecc533f8b8e57dc16cf4bc478615390e44a5f2ff1722747c74a475579148"},"source":{"id":"2605.06152","kind":"arxiv","version":3},"verdict":{"id":"73fe5ca6-aeea-4410-b502-b00c8227078a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-13T07:06:05.189778Z","strongest_claim":"This paper proves that this phenomenon is a result of floating-point arithmetic precision limits. ... We prove that this drift forms a positive feedback loop with the feature, causing the global classifier mean and the global feature mean to grow exponentially.","one_line_summary":"Slingshot loss spikes arise from floating-point precision limits that round correct-class gradients to zero, breaking zero-sum constraints and driving exponential parameter growth through numerical feature inflation.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The assumption that, once the logit difference exceeds the absorption-error threshold, the gradient of the correct class is rounded exactly to zero during backpropagation while incorrect-class gradients remain nonzero, and that this imbalance necessarily creates an exponential positive feedback loop with the features.","pith_extraction_headline":"Floating-point precision limits trigger slingshot loss spikes by creating numerical feature inflation in neural network training."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.06152/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"claim_evidence","ran_at":"2026-05-20T13:02:04.285171Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-20T08:36:17.189756Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_title_agreement","ran_at":"2026-05-19T19:01:19.348912Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T12:55:23.639732Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"5704dbfe57f64c954f96da93292ab57ce4e9d70270532fa950c2c4fa2712e08a"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b9b5ab6c7044a478fdc62f310b3c9f9d4878575de948cce219a613dc5ec39778"},"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":"2605.06152","created_at":"2026-05-27T01:05:56.307645+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.06152v3","created_at":"2026-05-27T01:05:56.307645+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.06152","created_at":"2026-05-27T01:05:56.307645+00:00"},{"alias_kind":"pith_short_12","alias_value":"L7NYTNSTZJXP","created_at":"2026-05-27T01:05:56.307645+00:00"},{"alias_kind":"pith_short_16","alias_value":"L7NYTNSTZJXPIILK","created_at":"2026-05-27T01:05:56.307645+00:00"},{"alias_kind":"pith_short_8","alias_value":"L7NYTNST","created_at":"2026-05-27T01:05:56.307645+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/L7NYTNSTZJXPIILKGOVVKEXZ54","json":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54.json","graph_json":"https://pith.science/api/pith-number/L7NYTNSTZJXPIILKGOVVKEXZ54/graph.json","events_json":"https://pith.science/api/pith-number/L7NYTNSTZJXPIILKGOVVKEXZ54/events.json","paper":"https://pith.science/paper/L7NYTNST"},"agent_actions":{"view_html":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54","download_json":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54.json","view_paper":"https://pith.science/paper/L7NYTNST","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.06152&json=true","fetch_graph":"https://pith.science/api/pith-number/L7NYTNSTZJXPIILKGOVVKEXZ54/graph.json","fetch_events":"https://pith.science/api/pith-number/L7NYTNSTZJXPIILKGOVVKEXZ54/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54/action/timestamp_anchor","attest_storage":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54/action/storage_attestation","attest_author":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54/action/author_attestation","sign_citation":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54/action/citation_signature","submit_replication":"https://pith.science/pith/L7NYTNSTZJXPIILKGOVVKEXZ54/action/replication_record"}},"created_at":"2026-05-27T01:05:56.307645+00:00","updated_at":"2026-05-27T01:05:56.307645+00:00"}