{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:5CX2PMOIDW63ZMBJWFAESW7CMN","short_pith_number":"pith:5CX2PMOI","schema_version":"1.0","canonical_sha256":"e8afa7b1c81dbdbcb029b140495be2635f51c109283d37a384a093559f3c6122","source":{"kind":"arxiv","id":"2406.02550","version":2},"attestation_state":"computed","paper":{"title":"Learning to grok: Emergence of in-context learning and skill composition in modular arithmetic tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","hep-th","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andrey Gromov, Aritra Das, Darshil Doshi, Tianyu He","submitted_at":"2024-06-04T17:59:36Z","abstract_excerpt":"Large language models can solve tasks that were not present in the training set. This capability is believed to be due to in-context learning and skill composition. In this work, we study the emergence of in-context learning and skill composition in a collection of modular arithmetic tasks. Specifically, we consider a finite collection of linear modular functions $z = a \\, x + b \\, y \\;\\mathrm{mod}\\; p$ labeled by the vector $(a, b) \\in \\mathbb{Z}_p^2$. We use some of these tasks for pre-training and the rest for out-of-distribution testing. We empirically show that a GPT-style transformer exh"},"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":"2406.02550","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2024-06-04T17:59:36Z","cross_cats_sorted":["cond-mat.dis-nn","hep-th","stat.ML"],"title_canon_sha256":"1080b689fe2dc7ca4201f0c673c361b73afc210ed890719ba484833bbe163bc8","abstract_canon_sha256":"ca296c1d3879ddcb7bee33ef475076974be11b627d7ef088e3772cbf6ecff4de"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:30:16.014848Z","signature_b64":"WQfGiOcxmgXuwt9Zy7U23sPUz4MQzaQBsHRWoC9hIeIYHTahQT2u7AufnszQQZW/14mDpEII8H01RW9N4AcZCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"e8afa7b1c81dbdbcb029b140495be2635f51c109283d37a384a093559f3c6122","last_reissued_at":"2026-07-05T09:30:16.014336Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:30:16.014336Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Learning to grok: Emergence of in-context learning and skill composition in modular arithmetic tasks","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cond-mat.dis-nn","hep-th","stat.ML"],"primary_cat":"cs.LG","authors_text":"Andrey Gromov, Aritra Das, Darshil Doshi, Tianyu He","submitted_at":"2024-06-04T17:59:36Z","abstract_excerpt":"Large language models can solve tasks that were not present in the training set. This capability is believed to be due to in-context learning and skill composition. In this work, we study the emergence of in-context learning and skill composition in a collection of modular arithmetic tasks. Specifically, we consider a finite collection of linear modular functions $z = a \\, x + b \\, y \\;\\mathrm{mod}\\; p$ labeled by the vector $(a, b) \\in \\mathbb{Z}_p^2$. We use some of these tasks for pre-training and the rest for out-of-distribution testing. We empirically show that a GPT-style transformer exh"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.02550","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/2406.02550/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":"2406.02550","created_at":"2026-07-05T09:30:16.014396+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.02550v2","created_at":"2026-07-05T09:30:16.014396+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.02550","created_at":"2026-07-05T09:30:16.014396+00:00"},{"alias_kind":"pith_short_12","alias_value":"5CX2PMOIDW63","created_at":"2026-07-05T09:30:16.014396+00:00"},{"alias_kind":"pith_short_16","alias_value":"5CX2PMOIDW63ZMBJ","created_at":"2026-07-05T09:30:16.014396+00:00"},{"alias_kind":"pith_short_8","alias_value":"5CX2PMOI","created_at":"2026-07-05T09:30:16.014396+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.12412","citing_title":"Stories in Space: In-Context Learning Trajectories in Conceptual Belief Space","ref_index":39,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09724","citing_title":"Model Capacity Determines Grokking through Competing Memorisation and Generalisation Speeds","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2605.05115","citing_title":"Manifold Steering Reveals the Shared Geometry of Neural Network Representation and Behavior","ref_index":267,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN","json":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN.json","graph_json":"https://pith.science/api/pith-number/5CX2PMOIDW63ZMBJWFAESW7CMN/graph.json","events_json":"https://pith.science/api/pith-number/5CX2PMOIDW63ZMBJWFAESW7CMN/events.json","paper":"https://pith.science/paper/5CX2PMOI"},"agent_actions":{"view_html":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN","download_json":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN.json","view_paper":"https://pith.science/paper/5CX2PMOI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.02550&json=true","fetch_graph":"https://pith.science/api/pith-number/5CX2PMOIDW63ZMBJWFAESW7CMN/graph.json","fetch_events":"https://pith.science/api/pith-number/5CX2PMOIDW63ZMBJWFAESW7CMN/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN/action/storage_attestation","attest_author":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN/action/author_attestation","sign_citation":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN/action/citation_signature","submit_replication":"https://pith.science/pith/5CX2PMOIDW63ZMBJWFAESW7CMN/action/replication_record"}},"created_at":"2026-07-05T09:30:16.014396+00:00","updated_at":"2026-07-05T09:30:16.014396+00:00"}