{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:6DM7LPVALM6XUHRVWPBV5L4Q27","short_pith_number":"pith:6DM7LPVA","schema_version":"1.0","canonical_sha256":"f0d9f5bea05b3d7a1e35b3c35eaf90d7ff5c85d45ddb3061772a3ff70444412f","source":{"kind":"arxiv","id":"2505.19713","version":3},"attestation_state":"computed","paper":{"title":"CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric Reward","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.GR","authors_text":"Dong Xu, Jing Zhang, Qian Yu, Xilin Wang, Ximing Xing, Yandong Guan","submitted_at":"2025-05-26T09:01:56Z","abstract_excerpt":"In this work, we introduce CAD-Coder, a novel framework that reformulates text-to-CAD as the generation of CadQuery scripts - a Python-based, parametric CAD language. This representation enables direct geometric validation, a richer modeling vocabulary, and seamless integration with existing LLMs. To further enhance code validity and geometric fidelity, we propose a two-stage learning pipeline: (1) supervised fine-tuning on paired text-CadQuery data, and (2) reinforcement learning with Group Reward Policy Optimization (GRPO), guided by a CAD-specific reward comprising both a geometric reward ("},"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":"2505.19713","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.GR","submitted_at":"2025-05-26T09:01:56Z","cross_cats_sorted":[],"title_canon_sha256":"ec4fd823990eed11282f4df8defc2d5cf6105665d4b1df6c629c56d66c58b3df","abstract_canon_sha256":"45f027b73d321055c159fbaf66b6ade5f62c5bf3ffcce422aaf61ecd31bdd9d2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:01.324270Z","signature_b64":"eoH9qycVZZxGnn8ArpojHGoQao2IyWB94a7ucAwr0izzLoIj1U0LOE8vtr9eW9d5hNPyoDm6xBSj4RCFhwMDCw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"f0d9f5bea05b3d7a1e35b3c35eaf90d7ff5c85d45ddb3061772a3ff70444412f","last_reissued_at":"2026-05-17T23:39:01.323628Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:01.323628Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CAD-Coder: Text-to-CAD Generation with Chain-of-Thought and Geometric Reward","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.GR","authors_text":"Dong Xu, Jing Zhang, Qian Yu, Xilin Wang, Ximing Xing, Yandong Guan","submitted_at":"2025-05-26T09:01:56Z","abstract_excerpt":"In this work, we introduce CAD-Coder, a novel framework that reformulates text-to-CAD as the generation of CadQuery scripts - a Python-based, parametric CAD language. This representation enables direct geometric validation, a richer modeling vocabulary, and seamless integration with existing LLMs. To further enhance code validity and geometric fidelity, we propose a two-stage learning pipeline: (1) supervised fine-tuning on paired text-CadQuery data, and (2) reinforcement learning with Group Reward Policy Optimization (GRPO), guided by a CAD-specific reward comprising both a geometric reward ("},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2505.19713","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"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":"2505.19713","created_at":"2026-05-17T23:39:01.323723+00:00"},{"alias_kind":"arxiv_version","alias_value":"2505.19713v3","created_at":"2026-05-17T23:39:01.323723+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2505.19713","created_at":"2026-05-17T23:39:01.323723+00:00"},{"alias_kind":"pith_short_12","alias_value":"6DM7LPVALM6X","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"6DM7LPVALM6XUHRV","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"6DM7LPVA","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2603.04337","citing_title":"Pointer-CAD: Unifying B-Rep and Command Sequences via Pointer-based Edges & Faces Selection","ref_index":17,"is_internal_anchor":true},{"citing_arxiv_id":"2604.03144","citing_title":"InCoder-32B-Thinking: Industrial Code World Model for Thinking","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2604.10992","citing_title":"ArtiCAD: Articulated CAD Assembly Design via Multi-Agent Code Generation","ref_index":17,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27","json":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27.json","graph_json":"https://pith.science/api/pith-number/6DM7LPVALM6XUHRVWPBV5L4Q27/graph.json","events_json":"https://pith.science/api/pith-number/6DM7LPVALM6XUHRVWPBV5L4Q27/events.json","paper":"https://pith.science/paper/6DM7LPVA"},"agent_actions":{"view_html":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27","download_json":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27.json","view_paper":"https://pith.science/paper/6DM7LPVA","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2505.19713&json=true","fetch_graph":"https://pith.science/api/pith-number/6DM7LPVALM6XUHRVWPBV5L4Q27/graph.json","fetch_events":"https://pith.science/api/pith-number/6DM7LPVALM6XUHRVWPBV5L4Q27/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27/action/timestamp_anchor","attest_storage":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27/action/storage_attestation","attest_author":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27/action/author_attestation","sign_citation":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27/action/citation_signature","submit_replication":"https://pith.science/pith/6DM7LPVALM6XUHRVWPBV5L4Q27/action/replication_record"}},"created_at":"2026-05-17T23:39:01.323723+00:00","updated_at":"2026-05-17T23:39:01.323723+00:00"}