{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:SI45DNWB3ZSRZMPXO64VIZATNO","short_pith_number":"pith:SI45DNWB","schema_version":"1.0","canonical_sha256":"9239d1b6c1de651cb1f777b95464136ba0ea5804fce6a55beb95dc5cbf8c76b2","source":{"kind":"arxiv","id":"2508.05242","version":1},"attestation_state":"computed","paper":{"title":"CodeBoost: Boosting Code LLMs by Squeezing Knowledge from Code Snippets with RL","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Kai Zhao, Quanjiang Guo, Rui She, Shangshu Yu, Sijie Wang, Siqi Li, Wee Peng Tay, Xiang Li, Xin Li, Yawei Zhang","submitted_at":"2025-08-07T10:31:24Z","abstract_excerpt":"Code large language models (LLMs) have become indispensable tools for building efficient and automated coding pipelines. Existing models are typically post-trained using reinforcement learning (RL) from general-purpose LLMs using \"human instruction-final answer\" pairs, where the instructions are usually from manual annotations. However, collecting high-quality coding instructions is both labor-intensive and difficult to scale. On the other hand, code snippets are abundantly available from various sources. This imbalance presents a major bottleneck in instruction-based post-training. We propose"},"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":"2508.05242","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2025-08-07T10:31:24Z","cross_cats_sorted":[],"title_canon_sha256":"c169074ebb67f2cb1bf50bb9c49bc60be35f83367719b8ecf8863ad5800c6bf7","abstract_canon_sha256":"4f38c01f0b170062895243d708a47784afd9d44cd537a1142e60c80683f13901"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:50:09.675256Z","signature_b64":"4gjjVWvK+sEvp4rdWwJY2fJSoKoCNs5ZMkwnnbk/C0ED08v9/8LFg8KYf6Ct3TMzSfrqX6oQo99uBgfCMYRRDA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9239d1b6c1de651cb1f777b95464136ba0ea5804fce6a55beb95dc5cbf8c76b2","last_reissued_at":"2026-07-05T11:50:09.674670Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:50:09.674670Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CodeBoost: Boosting Code LLMs by Squeezing Knowledge from Code Snippets with RL","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Kai Zhao, Quanjiang Guo, Rui She, Shangshu Yu, Sijie Wang, Siqi Li, Wee Peng Tay, Xiang Li, Xin Li, Yawei Zhang","submitted_at":"2025-08-07T10:31:24Z","abstract_excerpt":"Code large language models (LLMs) have become indispensable tools for building efficient and automated coding pipelines. Existing models are typically post-trained using reinforcement learning (RL) from general-purpose LLMs using \"human instruction-final answer\" pairs, where the instructions are usually from manual annotations. However, collecting high-quality coding instructions is both labor-intensive and difficult to scale. On the other hand, code snippets are abundantly available from various sources. This imbalance presents a major bottleneck in instruction-based post-training. We propose"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.05242","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/2508.05242/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":"2508.05242","created_at":"2026-07-05T11:50:09.674738+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.05242v1","created_at":"2026-07-05T11:50:09.674738+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.05242","created_at":"2026-07-05T11:50:09.674738+00:00"},{"alias_kind":"pith_short_12","alias_value":"SI45DNWB3ZSR","created_at":"2026-07-05T11:50:09.674738+00:00"},{"alias_kind":"pith_short_16","alias_value":"SI45DNWB3ZSRZMPX","created_at":"2026-07-05T11:50:09.674738+00:00"},{"alias_kind":"pith_short_8","alias_value":"SI45DNWB","created_at":"2026-07-05T11:50:09.674738+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":4,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.27733","citing_title":"BashCoder-R1: Towards Robust and Explainable Bash Code Generation with Robustness-Aware Group Relative Policy Optimization","ref_index":46,"is_internal_anchor":false},{"citing_arxiv_id":"2510.18471","citing_title":"CodeRL+: Improving Code Generation via Reinforcement with Execution Semantics Alignment","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2603.29957","citing_title":"Think Anywhere in Code Generation","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.00433","citing_title":"Improving LLM Code Generation via Requirement-Aware Curriculum Reinforcement Learning","ref_index":52,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO","json":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO.json","graph_json":"https://pith.science/api/pith-number/SI45DNWB3ZSRZMPXO64VIZATNO/graph.json","events_json":"https://pith.science/api/pith-number/SI45DNWB3ZSRZMPXO64VIZATNO/events.json","paper":"https://pith.science/paper/SI45DNWB"},"agent_actions":{"view_html":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO","download_json":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO.json","view_paper":"https://pith.science/paper/SI45DNWB","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.05242&json=true","fetch_graph":"https://pith.science/api/pith-number/SI45DNWB3ZSRZMPXO64VIZATNO/graph.json","fetch_events":"https://pith.science/api/pith-number/SI45DNWB3ZSRZMPXO64VIZATNO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO/action/storage_attestation","attest_author":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO/action/author_attestation","sign_citation":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO/action/citation_signature","submit_replication":"https://pith.science/pith/SI45DNWB3ZSRZMPXO64VIZATNO/action/replication_record"}},"created_at":"2026-07-05T11:50:09.674738+00:00","updated_at":"2026-07-05T11:50:09.674738+00:00"}