{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:TTCQHR3NLUMF52SF2ALTC2ACVC","short_pith_number":"pith:TTCQHR3N","schema_version":"1.0","canonical_sha256":"9cc503c76d5d185eea45d017316802a8ad33e80a2e9b9fd9dab7e1f8be4d12c0","source":{"kind":"arxiv","id":"2305.02309","version":2},"attestation_state":"computed","paper":{"title":"CodeGen2: Lessons for Training LLMs on Programming and Natural Languages","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Caiming Xiong, Erik Nijkamp, Hiroaki Hayashi, Silvio Savarese, Yingbo Zhou","submitted_at":"2023-05-03T17:55:25Z","abstract_excerpt":"Large language models (LLMs) have demonstrated remarkable abilities in representation learning for program synthesis and understanding tasks. The quality of the learned representations appears to be dictated by the neural scaling laws as a function of the number of model parameters and observations, while imposing upper bounds on the model performance by the amount of available data and compute, which is costly.\n  In this study, we attempt to render the training of LLMs for program synthesis more efficient by unifying four key components: (1) model architectures, (2) learning methods, (3) infi"},"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":"2305.02309","kind":"arxiv","version":2},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-05-03T17:55:25Z","cross_cats_sorted":[],"title_canon_sha256":"99216023c51470c39c66fdd1ec99265086d3d2b7c6bbd92637d54ddf6908d4a4","abstract_canon_sha256":"80139de5016cab931cca73d08ee7875dcba2ba2b668118a28c5eca72c84f0c0e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T06:30:06.818991Z","signature_b64":"Y1en7m8KZ+mETMWhQSec+y5pKrfT3JpjdS4CiYl3tQWbHHJjWdiCPUHYKf0y1iOjOMZclE8B6Ihap0orA4zBBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9cc503c76d5d185eea45d017316802a8ad33e80a2e9b9fd9dab7e1f8be4d12c0","last_reissued_at":"2026-07-05T06:30:06.818559Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T06:30:06.818559Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CodeGen2: Lessons for Training LLMs on Programming and Natural Languages","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Caiming Xiong, Erik Nijkamp, Hiroaki Hayashi, Silvio Savarese, Yingbo Zhou","submitted_at":"2023-05-03T17:55:25Z","abstract_excerpt":"Large language models (LLMs) have demonstrated remarkable abilities in representation learning for program synthesis and understanding tasks. The quality of the learned representations appears to be dictated by the neural scaling laws as a function of the number of model parameters and observations, while imposing upper bounds on the model performance by the amount of available data and compute, which is costly.\n  In this study, we attempt to render the training of LLMs for program synthesis more efficient by unifying four key components: (1) model architectures, (2) learning methods, (3) infi"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2305.02309","kind":"arxiv","version":2},"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/2305.02309/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":"2305.02309","created_at":"2026-07-05T06:30:06.818616+00:00"},{"alias_kind":"arxiv_version","alias_value":"2305.02309v2","created_at":"2026-07-05T06:30:06.818616+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2305.02309","created_at":"2026-07-05T06:30:06.818616+00:00"},{"alias_kind":"pith_short_12","alias_value":"TTCQHR3NLUMF","created_at":"2026-07-05T06:30:06.818616+00:00"},{"alias_kind":"pith_short_16","alias_value":"TTCQHR3NLUMF52SF","created_at":"2026-07-05T06:30:06.818616+00:00"},{"alias_kind":"pith_short_8","alias_value":"TTCQHR3N","created_at":"2026-07-05T06:30:06.818616+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":14,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2607.01883","citing_title":"PairCoder++: Pair Programming as a Universal Paradigm for Verified Code-Driven Multimodal and Structured-Artifact Generation","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2606.30810","citing_title":"Towards Knowledge Alignment in Code LLMs: Contrastive Unlearning for Evolving APIs","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2507.00642","citing_title":"ChatHLS: Towards Systematic Design Automation and Optimization for High-Level Synthesis","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2511.01763","citing_title":"Context-Guided Decompilation: A Step Towards Re-executability","ref_index":43,"is_internal_anchor":false},{"citing_arxiv_id":"2406.00515","citing_title":"A Survey on Large Language Models for Code Generation","ref_index":196,"is_internal_anchor":false},{"citing_arxiv_id":"2306.11644","citing_title":"Textbooks Are All You Need","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2305.15334","citing_title":"Gorilla: Large Language Model Connected with Massive APIs","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2604.00239","citing_title":"A Taxonomy of Programming Languages for Code Generation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2402.06196","citing_title":"Large Language Models: A Survey","ref_index":113,"is_internal_anchor":false},{"citing_arxiv_id":"2604.18162","citing_title":"VerilogCL: A Contrastive Learning Framework for Robust LLM-Based Verilog Generation","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04066","citing_title":"Adapt to Thrive! Adaptive Power-Mean Policy Optimization for Improved LLM Reasoning","ref_index":212,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04065","citing_title":"Free Energy-Driven Reinforcement Learning with Adaptive Advantage Shaping for Unsupervised Reasoning in LLMs","ref_index":227,"is_internal_anchor":false},{"citing_arxiv_id":"2303.18223","citing_title":"A Survey of Large Language Models","ref_index":99,"is_internal_anchor":false},{"citing_arxiv_id":"2604.17184","citing_title":"SynthFix: Adaptive Neuro-Symbolic Code Vulnerability Repair","ref_index":23,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC","json":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC.json","graph_json":"https://pith.science/api/pith-number/TTCQHR3NLUMF52SF2ALTC2ACVC/graph.json","events_json":"https://pith.science/api/pith-number/TTCQHR3NLUMF52SF2ALTC2ACVC/events.json","paper":"https://pith.science/paper/TTCQHR3N"},"agent_actions":{"view_html":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC","download_json":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC.json","view_paper":"https://pith.science/paper/TTCQHR3N","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2305.02309&json=true","fetch_graph":"https://pith.science/api/pith-number/TTCQHR3NLUMF52SF2ALTC2ACVC/graph.json","fetch_events":"https://pith.science/api/pith-number/TTCQHR3NLUMF52SF2ALTC2ACVC/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC/action/timestamp_anchor","attest_storage":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC/action/storage_attestation","attest_author":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC/action/author_attestation","sign_citation":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC/action/citation_signature","submit_replication":"https://pith.science/pith/TTCQHR3NLUMF52SF2ALTC2ACVC/action/replication_record"}},"created_at":"2026-07-05T06:30:06.818616+00:00","updated_at":"2026-07-05T06:30:06.818616+00:00"}