{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:7MXNVDTEMB3TGTXXUIWYHM757X","short_pith_number":"pith:7MXNVDTE","schema_version":"1.0","canonical_sha256":"fb2eda8e646077334ef7a22d83b3fdfddc09c5a97cb0ac1101b8965275c55af1","source":{"kind":"arxiv","id":"2402.04875","version":6},"attestation_state":"computed","paper":{"title":"On Provable Length and Compositional Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Amin Mansouri, Kartik Ahuja","submitted_at":"2024-02-07T14:16:28Z","abstract_excerpt":"Out-of-distribution generalization capabilities of sequence-to-sequence models can be studied from the lens of two crucial forms of generalization: length generalization -- the ability to generalize to longer sequences than ones seen during training, and compositional generalization: the ability to generalize to token combinations not seen during training. In this work, we provide first provable guarantees on length and compositional generalization for common sequence-to-sequence models -- deep sets, transformers, state space models, and recurrent neural nets -- trained to minimize the predict"},"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":"2402.04875","kind":"arxiv","version":6},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2024-02-07T14:16:28Z","cross_cats_sorted":["cs.CL","stat.ML"],"title_canon_sha256":"2afaa4c71d7259b534c7ba16352aabad40763b4e1d30286cf0c24be073e3a4bf","abstract_canon_sha256":"b74c318af002200ad85ee589f5fcacef2944684fdede44caa6a28025952daa0d"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:10:45.026056Z","signature_b64":"y0/GDVp9GogDhY/8aVFBg+cEtXKlKv1OtE9P9FQQFdbR1EfO/xjBgVu+XNv072FJIbritZe3jL9uKZAHqA6GAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"fb2eda8e646077334ef7a22d83b3fdfddc09c5a97cb0ac1101b8965275c55af1","last_reissued_at":"2026-07-05T11:10:45.025539Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:10:45.025539Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"On Provable Length and Compositional Generalization","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CL","stat.ML"],"primary_cat":"cs.LG","authors_text":"Amin Mansouri, Kartik Ahuja","submitted_at":"2024-02-07T14:16:28Z","abstract_excerpt":"Out-of-distribution generalization capabilities of sequence-to-sequence models can be studied from the lens of two crucial forms of generalization: length generalization -- the ability to generalize to longer sequences than ones seen during training, and compositional generalization: the ability to generalize to token combinations not seen during training. In this work, we provide first provable guarantees on length and compositional generalization for common sequence-to-sequence models -- deep sets, transformers, state space models, and recurrent neural nets -- trained to minimize the predict"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2402.04875","kind":"arxiv","version":6},"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/2402.04875/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":"2402.04875","created_at":"2026-07-05T11:10:45.025597+00:00"},{"alias_kind":"arxiv_version","alias_value":"2402.04875v6","created_at":"2026-07-05T11:10:45.025597+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2402.04875","created_at":"2026-07-05T11:10:45.025597+00:00"},{"alias_kind":"pith_short_12","alias_value":"7MXNVDTEMB3T","created_at":"2026-07-05T11:10:45.025597+00:00"},{"alias_kind":"pith_short_16","alias_value":"7MXNVDTEMB3TGTXX","created_at":"2026-07-05T11:10:45.025597+00:00"},{"alias_kind":"pith_short_8","alias_value":"7MXNVDTE","created_at":"2026-07-05T11:10:45.025597+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06720","citing_title":"When Does In-Context Search Help? A Sampling-Complexity Theory of Reflection-Driven Reasoning","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2606.00183","citing_title":"Agentic Transformers Provably Learn to Search via Reinforcement Learning","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2604.15306","citing_title":"Generalization in LLM Problem Solving: The Case of the Shortest Path","ref_index":2,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X","json":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X.json","graph_json":"https://pith.science/api/pith-number/7MXNVDTEMB3TGTXXUIWYHM757X/graph.json","events_json":"https://pith.science/api/pith-number/7MXNVDTEMB3TGTXXUIWYHM757X/events.json","paper":"https://pith.science/paper/7MXNVDTE"},"agent_actions":{"view_html":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X","download_json":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X.json","view_paper":"https://pith.science/paper/7MXNVDTE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2402.04875&json=true","fetch_graph":"https://pith.science/api/pith-number/7MXNVDTEMB3TGTXXUIWYHM757X/graph.json","fetch_events":"https://pith.science/api/pith-number/7MXNVDTEMB3TGTXXUIWYHM757X/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X/action/timestamp_anchor","attest_storage":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X/action/storage_attestation","attest_author":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X/action/author_attestation","sign_citation":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X/action/citation_signature","submit_replication":"https://pith.science/pith/7MXNVDTEMB3TGTXXUIWYHM757X/action/replication_record"}},"created_at":"2026-07-05T11:10:45.025597+00:00","updated_at":"2026-07-05T11:10:45.025597+00:00"}