{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:HVWT2AGUB6OLPX45H75AX3W526","short_pith_number":"pith:HVWT2AGU","schema_version":"1.0","canonical_sha256":"3d6d3d00d40f9cb7df9d3ffa0beeddd7a2e424b2307e18fe042da32c448b326a","source":{"kind":"arxiv","id":"2504.12687","version":1},"attestation_state":"computed","paper":{"title":"Data-efficient LLM Fine-tuning for Code Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Sheng-Jun Huang, Weijie Lv, Xuan Xia","submitted_at":"2025-04-17T06:29:28Z","abstract_excerpt":"Large language models (LLMs) have demonstrated significant potential in code generation tasks. However, there remains a performance gap between open-source and closed-source models. To address this gap, existing approaches typically generate large amounts of synthetic data for fine-tuning, which often leads to inefficient training. In this work, we propose a data selection strategy in order to improve the effectiveness and efficiency of training for code-based LLMs. By prioritizing data complexity and ensuring that the sampled subset aligns with the distribution of the original dataset, our sa"},"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":"2504.12687","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-04-17T06:29:28Z","cross_cats_sorted":[],"title_canon_sha256":"d8375f47dea016117bdf21589f8097376ce6e400c3b8117a2b402d0caba89352","abstract_canon_sha256":"ce67d1a45b15a860e14c7cb4dfc7f426f883d41bbb25bf9072779c8ab1558c90"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:50:20.632411Z","signature_b64":"uEKDM6LSf1ocQhBr/BS5KmQrrcKu3uKzKTOlYvkDUBn1MsqQ/4ftFEXtUdUzbKmlHmqpj2vSKZkM9TPuVoMfCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"3d6d3d00d40f9cb7df9d3ffa0beeddd7a2e424b2307e18fe042da32c448b326a","last_reissued_at":"2026-07-05T10:50:20.631888Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:50:20.631888Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Data-efficient LLM Fine-tuning for Code Generation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Sheng-Jun Huang, Weijie Lv, Xuan Xia","submitted_at":"2025-04-17T06:29:28Z","abstract_excerpt":"Large language models (LLMs) have demonstrated significant potential in code generation tasks. However, there remains a performance gap between open-source and closed-source models. To address this gap, existing approaches typically generate large amounts of synthetic data for fine-tuning, which often leads to inefficient training. In this work, we propose a data selection strategy in order to improve the effectiveness and efficiency of training for code-based LLMs. By prioritizing data complexity and ensuring that the sampled subset aligns with the distribution of the original dataset, our sa"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2504.12687","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/2504.12687/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":"2504.12687","created_at":"2026-07-05T10:50:20.631952+00:00"},{"alias_kind":"arxiv_version","alias_value":"2504.12687v1","created_at":"2026-07-05T10:50:20.631952+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2504.12687","created_at":"2026-07-05T10:50:20.631952+00:00"},{"alias_kind":"pith_short_12","alias_value":"HVWT2AGUB6OL","created_at":"2026-07-05T10:50:20.631952+00:00"},{"alias_kind":"pith_short_16","alias_value":"HVWT2AGUB6OLPX45","created_at":"2026-07-05T10:50:20.631952+00:00"},{"alias_kind":"pith_short_8","alias_value":"HVWT2AGU","created_at":"2026-07-05T10:50:20.631952+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.05267","citing_title":"Bridging Generation and Training: A Systematic Review of Quality Issues in LLMs for Code","ref_index":82,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526","json":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526.json","graph_json":"https://pith.science/api/pith-number/HVWT2AGUB6OLPX45H75AX3W526/graph.json","events_json":"https://pith.science/api/pith-number/HVWT2AGUB6OLPX45H75AX3W526/events.json","paper":"https://pith.science/paper/HVWT2AGU"},"agent_actions":{"view_html":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526","download_json":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526.json","view_paper":"https://pith.science/paper/HVWT2AGU","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2504.12687&json=true","fetch_graph":"https://pith.science/api/pith-number/HVWT2AGUB6OLPX45H75AX3W526/graph.json","fetch_events":"https://pith.science/api/pith-number/HVWT2AGUB6OLPX45H75AX3W526/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526/action/timestamp_anchor","attest_storage":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526/action/storage_attestation","attest_author":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526/action/author_attestation","sign_citation":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526/action/citation_signature","submit_replication":"https://pith.science/pith/HVWT2AGUB6OLPX45H75AX3W526/action/replication_record"}},"created_at":"2026-07-05T10:50:20.631952+00:00","updated_at":"2026-07-05T10:50:20.631952+00:00"}