{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:YI7EMO4V3YCFL4HTNT7ZMT4SSR","short_pith_number":"pith:YI7EMO4V","schema_version":"1.0","canonical_sha256":"c23e463b95de0455f0f36cff964f92944b022f21a7c967d576c5f2292a072173","source":{"kind":"arxiv","id":"2607.02390","version":1},"attestation_state":"computed","paper":{"title":"DecompRL: Solving Harder Problems by Learning Modular Code Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Fabian Gloeckle, Francis Bach, Gabriel Synnaeve, Juliette Decugis, Taco Cohen","submitted_at":"2026-07-02T16:25:10Z","abstract_excerpt":"How can Large Language Models (LLMs) solve problems they currently cannot? Repeated sampling scales test-time compute but GPU cost grows linearly with attempts, while reinforcement learning (RL) with verifiable rewards improves single-attempt accuracy at the expense of sample diversity. Both strategies ultimately fail when the base policy has near-zero probability of producing a correct solution: no amount of sampling or gradient signal can overcome a search space that is simply too large. We take a different approach: rather than sampling harder, we make the task easier by decomposing problem"},"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":"2607.02390","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2026-07-02T16:25:10Z","cross_cats_sorted":[],"title_canon_sha256":"7bbd77dff77b93f74d300828db50b1725034e8a4bc2394e502f7d34fc98c3ffd","abstract_canon_sha256":"1771422e9672a21205ce089cb8ab03291e6899bbb406f74088f486201374c6ba"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-03T01:17:57.467063Z","signature_b64":"kvVpZnP/3M1j8/PGFm2vwlN046gbuVSW6SUkdZ71jvsZgtrpPM0yS2Y8XXxT4m9kCDisM/KIKJ1DXZ4JQa+YAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"c23e463b95de0455f0f36cff964f92944b022f21a7c967d576c5f2292a072173","last_reissued_at":"2026-07-03T01:17:57.466689Z","signature_status":"signed_v1","first_computed_at":"2026-07-03T01:17:57.466689Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DecompRL: Solving Harder Problems by Learning Modular Code Generation","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Fabian Gloeckle, Francis Bach, Gabriel Synnaeve, Juliette Decugis, Taco Cohen","submitted_at":"2026-07-02T16:25:10Z","abstract_excerpt":"How can Large Language Models (LLMs) solve problems they currently cannot? Repeated sampling scales test-time compute but GPU cost grows linearly with attempts, while reinforcement learning (RL) with verifiable rewards improves single-attempt accuracy at the expense of sample diversity. Both strategies ultimately fail when the base policy has near-zero probability of producing a correct solution: no amount of sampling or gradient signal can overcome a search space that is simply too large. We take a different approach: rather than sampling harder, we make the task easier by decomposing problem"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.02390","kind":"arxiv","version":1},"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/2607.02390/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":"2607.02390","created_at":"2026-07-03T01:17:57.466750+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.02390v1","created_at":"2026-07-03T01:17:57.466750+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.02390","created_at":"2026-07-03T01:17:57.466750+00:00"},{"alias_kind":"pith_short_12","alias_value":"YI7EMO4V3YCF","created_at":"2026-07-03T01:17:57.466750+00:00"},{"alias_kind":"pith_short_16","alias_value":"YI7EMO4V3YCFL4HT","created_at":"2026-07-03T01:17:57.466750+00:00"},{"alias_kind":"pith_short_8","alias_value":"YI7EMO4V","created_at":"2026-07-03T01:17:57.466750+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR","json":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR.json","graph_json":"https://pith.science/api/pith-number/YI7EMO4V3YCFL4HTNT7ZMT4SSR/graph.json","events_json":"https://pith.science/api/pith-number/YI7EMO4V3YCFL4HTNT7ZMT4SSR/events.json","paper":"https://pith.science/paper/YI7EMO4V"},"agent_actions":{"view_html":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR","download_json":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR.json","view_paper":"https://pith.science/paper/YI7EMO4V","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.02390&json=true","fetch_graph":"https://pith.science/api/pith-number/YI7EMO4V3YCFL4HTNT7ZMT4SSR/graph.json","fetch_events":"https://pith.science/api/pith-number/YI7EMO4V3YCFL4HTNT7ZMT4SSR/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR/action/timestamp_anchor","attest_storage":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR/action/storage_attestation","attest_author":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR/action/author_attestation","sign_citation":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR/action/citation_signature","submit_replication":"https://pith.science/pith/YI7EMO4V3YCFL4HTNT7ZMT4SSR/action/replication_record"}},"created_at":"2026-07-03T01:17:57.466750+00:00","updated_at":"2026-07-03T01:17:57.466750+00:00"}