{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:NRY64HA6SMMAKHEKMWL6NMZZZC","short_pith_number":"pith:NRY64HA6","schema_version":"1.0","canonical_sha256":"6c71ee1c1e9318051c8a6597e6b339c88273c5e9648db6511d40cf9a697c7546","source":{"kind":"arxiv","id":"2410.11778","version":3},"attestation_state":"computed","paper":{"title":"On the Training Convergence of Transformers for In-Context Classification of Gaussian Mixtures","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IT","math.IT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Cong Shen, Jing Yang, Ruida Zhou, Wei Shen","submitted_at":"2024-10-15T16:57:14Z","abstract_excerpt":"Although transformers have demonstrated impressive capabilities for in-context learning (ICL) in practice, theoretical understanding of the underlying mechanism that allows transformers to perform ICL is still in its infancy. This work aims to theoretically study the training dynamics of transformers for in-context classification tasks. We demonstrate that, for in-context classification of Gaussian mixtures under certain assumptions, a single-layer transformer trained via gradient descent converges to a globally optimal model at a linear rate. We further quantify the impact of the training and"},"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":"2410.11778","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-10-15T16:57:14Z","cross_cats_sorted":["cs.IT","math.IT","stat.ML"],"title_canon_sha256":"1b208a204072e0eddc021ddb0338e7875a2cb59d15abf43a8d9194710fdc1329","abstract_canon_sha256":"0fb668b76ba7babac356fa30e4ca63ed633dc54ce6dd2e000eded0b56828be06"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:12:10.966591Z","signature_b64":"pIm6EkFL77we3f720HBckQmHvVD87f+V+kD5qLauOJFzmsfAA99vI2XOjxDtXd/KE87yV+OwM/g/75/1BM7FBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6c71ee1c1e9318051c8a6597e6b339c88273c5e9648db6511d40cf9a697c7546","last_reissued_at":"2026-07-05T11:12:10.966081Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:12:10.966081Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On the Training Convergence of Transformers for In-Context Classification of Gaussian Mixtures","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.IT","math.IT","stat.ML"],"primary_cat":"cs.LG","authors_text":"Cong Shen, Jing Yang, Ruida Zhou, Wei Shen","submitted_at":"2024-10-15T16:57:14Z","abstract_excerpt":"Although transformers have demonstrated impressive capabilities for in-context learning (ICL) in practice, theoretical understanding of the underlying mechanism that allows transformers to perform ICL is still in its infancy. This work aims to theoretically study the training dynamics of transformers for in-context classification tasks. We demonstrate that, for in-context classification of Gaussian mixtures under certain assumptions, a single-layer transformer trained via gradient descent converges to a globally optimal model at a linear rate. We further quantify the impact of the training and"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.11778","kind":"arxiv","version":3},"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/2410.11778/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":"2410.11778","created_at":"2026-07-05T11:12:10.966143+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.11778v3","created_at":"2026-07-05T11:12:10.966143+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.11778","created_at":"2026-07-05T11:12:10.966143+00:00"},{"alias_kind":"pith_short_12","alias_value":"NRY64HA6SMMA","created_at":"2026-07-05T11:12:10.966143+00:00"},{"alias_kind":"pith_short_16","alias_value":"NRY64HA6SMMAKHEK","created_at":"2026-07-05T11:12:10.966143+00:00"},{"alias_kind":"pith_short_8","alias_value":"NRY64HA6","created_at":"2026-07-05T11:12:10.966143+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.30323","citing_title":"In-Context Reward Adaptation for Robust Preference Modeling","ref_index":12,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC","json":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC.json","graph_json":"https://pith.science/api/pith-number/NRY64HA6SMMAKHEKMWL6NMZZZC/graph.json","events_json":"https://pith.science/api/pith-number/NRY64HA6SMMAKHEKMWL6NMZZZC/events.json","paper":"https://pith.science/paper/NRY64HA6"},"agent_actions":{"view_html":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC","download_json":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC.json","view_paper":"https://pith.science/paper/NRY64HA6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.11778&json=true","fetch_graph":"https://pith.science/api/pith-number/NRY64HA6SMMAKHEKMWL6NMZZZC/graph.json","fetch_events":"https://pith.science/api/pith-number/NRY64HA6SMMAKHEKMWL6NMZZZC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC/action/storage_attestation","attest_author":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC/action/author_attestation","sign_citation":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC/action/citation_signature","submit_replication":"https://pith.science/pith/NRY64HA6SMMAKHEKMWL6NMZZZC/action/replication_record"}},"created_at":"2026-07-05T11:12:10.966143+00:00","updated_at":"2026-07-05T11:12:10.966143+00:00"}