{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:XJO22QNPIQFHRKQGBVWFFLHXYV","short_pith_number":"pith:XJO22QNP","schema_version":"1.0","canonical_sha256":"ba5dad41af440a78aa060d6c52acf7c57593aca822f8388584e6a95162f2aca4","source":{"kind":"arxiv","id":"2406.20060","version":1},"attestation_state":"computed","paper":{"title":"Applying RLAIF for Code Generation with API-usage in Lightweight LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bortik Bandyopadhyay, Raviteja Anantha, Sayantan Mahinder, Sujan Dutta","submitted_at":"2024-06-28T17:16:03Z","abstract_excerpt":"Reinforcement Learning from AI Feedback (RLAIF) has demonstrated significant potential across various domains, including mitigating harm in LLM outputs, enhancing text summarization, and mathematical reasoning. This paper introduces an RLAIF framework for improving the code generation abilities of lightweight (<1B parameters) LLMs. We specifically focus on code generation tasks that require writing appropriate API calls, which is challenging due to the well-known issue of hallucination in LLMs. Our framework extracts AI feedback from a larger LLM (e.g., GPT-3.5) through a specialized prompting"},"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":"2406.20060","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CL","submitted_at":"2024-06-28T17:16:03Z","cross_cats_sorted":[],"title_canon_sha256":"09e0e322dbe1144fc15e8fa09eb83c61da07f15b7c54953975ece3127985ebac","abstract_canon_sha256":"bd30b7e7770596083c577351b1b819515fca1744389c34fcdc188f56d065b9c4"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T08:37:58.538912Z","signature_b64":"z7DXIKhymHGxe+Q0klAXe90IPWSvcAeCGaiirXnQ+ntDkofEtL6uvbg+JjbuyUhjiikCJX+kr+Iy6TSVwD/wCA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ba5dad41af440a78aa060d6c52acf7c57593aca822f8388584e6a95162f2aca4","last_reissued_at":"2026-07-05T08:37:58.538417Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T08:37:58.538417Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Applying RLAIF for Code Generation with API-usage in Lightweight LLMs","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Bortik Bandyopadhyay, Raviteja Anantha, Sayantan Mahinder, Sujan Dutta","submitted_at":"2024-06-28T17:16:03Z","abstract_excerpt":"Reinforcement Learning from AI Feedback (RLAIF) has demonstrated significant potential across various domains, including mitigating harm in LLM outputs, enhancing text summarization, and mathematical reasoning. This paper introduces an RLAIF framework for improving the code generation abilities of lightweight (<1B parameters) LLMs. We specifically focus on code generation tasks that require writing appropriate API calls, which is challenging due to the well-known issue of hallucination in LLMs. Our framework extracts AI feedback from a larger LLM (e.g., GPT-3.5) through a specialized prompting"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.20060","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/2406.20060/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":"2406.20060","created_at":"2026-07-05T08:37:58.538476+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.20060v1","created_at":"2026-07-05T08:37:58.538476+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.20060","created_at":"2026-07-05T08:37:58.538476+00:00"},{"alias_kind":"pith_short_12","alias_value":"XJO22QNPIQFH","created_at":"2026-07-05T08:37:58.538476+00:00"},{"alias_kind":"pith_short_16","alias_value":"XJO22QNPIQFHRKQG","created_at":"2026-07-05T08:37:58.538476+00:00"},{"alias_kind":"pith_short_8","alias_value":"XJO22QNP","created_at":"2026-07-05T08:37:58.538476+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.05936","citing_title":"Mitigating Errors in LLM-Generated Web API Invocations via Retrieval-Augmented Generation and Constrained Decoding","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2605.07024","citing_title":"Delulu: A Verified Multi-Lingual Benchmark for Code Hallucination Detection in Fill-in-the-Middle Tasks","ref_index":3,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV","json":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV.json","graph_json":"https://pith.science/api/pith-number/XJO22QNPIQFHRKQGBVWFFLHXYV/graph.json","events_json":"https://pith.science/api/pith-number/XJO22QNPIQFHRKQGBVWFFLHXYV/events.json","paper":"https://pith.science/paper/XJO22QNP"},"agent_actions":{"view_html":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV","download_json":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV.json","view_paper":"https://pith.science/paper/XJO22QNP","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.20060&json=true","fetch_graph":"https://pith.science/api/pith-number/XJO22QNPIQFHRKQGBVWFFLHXYV/graph.json","fetch_events":"https://pith.science/api/pith-number/XJO22QNPIQFHRKQGBVWFFLHXYV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV/action/storage_attestation","attest_author":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV/action/author_attestation","sign_citation":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV/action/citation_signature","submit_replication":"https://pith.science/pith/XJO22QNPIQFHRKQGBVWFFLHXYV/action/replication_record"}},"created_at":"2026-07-05T08:37:58.538476+00:00","updated_at":"2026-07-05T08:37:58.538476+00:00"}