{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:IYQSZFQLVSZ6OGDF722PA2LY4V","short_pith_number":"pith:IYQSZFQL","schema_version":"1.0","canonical_sha256":"46212c960bacb3e71865feb4f06978e5666d8cc2a5a3c3436da98804bb99e2fb","source":{"kind":"arxiv","id":"2410.11268","version":2},"attestation_state":"computed","paper":{"title":"Bypassing the Exponential Dependency: Looped Transformers Efficiently Learn In-context by Multi-step Gradient Descent","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Bo Chen, Xiaoyu Li, Yingyu Liang, Zhao Song, Zhenmei Shi","submitted_at":"2024-10-15T04:44:23Z","abstract_excerpt":"In-context learning has been recognized as a key factor in the success of Large Language Models (LLMs). It refers to the model's ability to learn patterns on the fly from provided in-context examples in the prompt during inference. Previous studies have demonstrated that the Transformer architecture used in LLMs can implement a single-step gradient descent update by processing in-context examples in a single forward pass. Recent work has further shown that, during in-context learning, a looped Transformer can implement multi-step gradient descent updates in forward passes. However, their theor"},"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":"2410.11268","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-10-15T04:44:23Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"2ec93615c2e06b63a62ee668577f28a1302958b26016d5d7825df10d6b9b89ce","abstract_canon_sha256":"da56c2a355c745e30233c985d60c85eda9f83557948dd3187cd77e31abea7c62"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:22:07.814064Z","signature_b64":"abJB97qJvFcnSBpIKvQ78N2izKpTWCD2WJUV+Puf9DhaAn0emebbRZhn3AeNHsz1SbcteCuwmZHRvneZs7fZCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"46212c960bacb3e71865feb4f06978e5666d8cc2a5a3c3436da98804bb99e2fb","last_reissued_at":"2026-07-05T10:22:07.813391Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:22:07.813391Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Bypassing the Exponential Dependency: Looped Transformers Efficiently Learn In-context by Multi-step Gradient Descent","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Bo Chen, Xiaoyu Li, Yingyu Liang, Zhao Song, Zhenmei Shi","submitted_at":"2024-10-15T04:44:23Z","abstract_excerpt":"In-context learning has been recognized as a key factor in the success of Large Language Models (LLMs). It refers to the model's ability to learn patterns on the fly from provided in-context examples in the prompt during inference. Previous studies have demonstrated that the Transformer architecture used in LLMs can implement a single-step gradient descent update by processing in-context examples in a single forward pass. Recent work has further shown that, during in-context learning, a looped Transformer can implement multi-step gradient descent updates in forward passes. However, their theor"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2410.11268","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/2410.11268/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":"2410.11268","created_at":"2026-07-05T10:22:07.813476+00:00"},{"alias_kind":"arxiv_version","alias_value":"2410.11268v2","created_at":"2026-07-05T10:22:07.813476+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2410.11268","created_at":"2026-07-05T10:22:07.813476+00:00"},{"alias_kind":"pith_short_12","alias_value":"IYQSZFQLVSZ6","created_at":"2026-07-05T10:22:07.813476+00:00"},{"alias_kind":"pith_short_16","alias_value":"IYQSZFQLVSZ6OGDF","created_at":"2026-07-05T10:22:07.813476+00:00"},{"alias_kind":"pith_short_8","alias_value":"IYQSZFQL","created_at":"2026-07-05T10:22:07.813476+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.18797","citing_title":"Simply Stabilizing the Loop via Fully Looped Transformer","ref_index":10,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V","json":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V.json","graph_json":"https://pith.science/api/pith-number/IYQSZFQLVSZ6OGDF722PA2LY4V/graph.json","events_json":"https://pith.science/api/pith-number/IYQSZFQLVSZ6OGDF722PA2LY4V/events.json","paper":"https://pith.science/paper/IYQSZFQL"},"agent_actions":{"view_html":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V","download_json":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V.json","view_paper":"https://pith.science/paper/IYQSZFQL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2410.11268&json=true","fetch_graph":"https://pith.science/api/pith-number/IYQSZFQLVSZ6OGDF722PA2LY4V/graph.json","fetch_events":"https://pith.science/api/pith-number/IYQSZFQLVSZ6OGDF722PA2LY4V/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V/action/timestamp_anchor","attest_storage":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V/action/storage_attestation","attest_author":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V/action/author_attestation","sign_citation":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V/action/citation_signature","submit_replication":"https://pith.science/pith/IYQSZFQLVSZ6OGDF722PA2LY4V/action/replication_record"}},"created_at":"2026-07-05T10:22:07.813476+00:00","updated_at":"2026-07-05T10:22:07.813476+00:00"}