{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:EKHY7K6GY6BPHII4RY2PZL2LKY","short_pith_number":"pith:EKHY7K6G","canonical_record":{"source":{"id":"2604.10095","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-04-11T08:33:13Z","cross_cats_sorted":[],"title_canon_sha256":"05fb92cfbed80391deea7759521dcf3128e04b590a0ae7b55794cf2820d9a37d","abstract_canon_sha256":"a64f7a081ae6aefb495c6ccece4d3b026b2afac3cc4f2620a1c107f127e204f3"},"schema_version":"1.0"},"canonical_sha256":"228f8fabc6c782f3a11c8e34fcaf4b5630469a8d59e8678273219d4700351c51","source":{"kind":"arxiv","id":"2604.10095","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.10095","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"arxiv_version","alias_value":"2604.10095v2","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.10095","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_12","alias_value":"EKHY7K6GY6BP","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_16","alias_value":"EKHY7K6GY6BPHII4","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_8","alias_value":"EKHY7K6G","created_at":"2026-05-27T01:05:54Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:EKHY7K6GY6BPHII4RY2PZL2LKY","target":"record","payload":{"canonical_record":{"source":{"id":"2604.10095","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-04-11T08:33:13Z","cross_cats_sorted":[],"title_canon_sha256":"05fb92cfbed80391deea7759521dcf3128e04b590a0ae7b55794cf2820d9a37d","abstract_canon_sha256":"a64f7a081ae6aefb495c6ccece4d3b026b2afac3cc4f2620a1c107f127e204f3"},"schema_version":"1.0"},"canonical_sha256":"228f8fabc6c782f3a11c8e34fcaf4b5630469a8d59e8678273219d4700351c51","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T01:05:54.208473Z","signature_b64":"MHLajTpRMZuOZM+qna/TIg0/nzuT1cQw4sgwllKkaIUKR65Bekf5VVMHE/MZ9/WVtLAVkHqtZrDQxRn86sYPAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"228f8fabc6c782f3a11c8e34fcaf4b5630469a8d59e8678273219d4700351c51","last_reissued_at":"2026-05-27T01:05:54.207622Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T01:05:54.207622Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2604.10095","source_version":2,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-27T01:05:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"17NDsF7Hn/lXeMxf40VRdBxcNqIVK0dvTYYDF7NGXvJSwEG0ONnxsp+ICOjr25140cNVMHR0tTyiWvWRvdO+Bw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T03:40:15.994024Z"},"content_sha256":"241d7bc7a8a4ff7e40cc79c78ce15d734ba66c1987811240413ecf0a2504d6cc","schema_version":"1.0","event_id":"sha256:241d7bc7a8a4ff7e40cc79c78ce15d734ba66c1987811240413ecf0a2504d6cc"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:EKHY7K6GY6BPHII4RY2PZL2LKY","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Mining Attribute Subspaces for Efficient Fine-tuning of 3D Foundation Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Subspaces extracted from synthetic 3D variations combine into a compact LoRA adapter that raises fine-tuning accuracy on real data","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Ahmed Abdelkader, Brandon Y. Feng, Hanwen Jiang, Qixing Huang, Wen-Sheng Chu, Yu Jiang, Zhangyang Wang","submitted_at":"2026-04-11T08:33:13Z","abstract_excerpt":"With the emergence of 3D foundation models, there is growing interest in fine-tuning them for downstream tasks, where LoRA is the dominant fine-tuning paradigm. As 3D datasets exhibit distinct variations in texture, geometry, camera motion, and lighting, there are interesting fundamental questions: 1) Are there LoRA subspaces associated with each type of variation? 2) Are these subspaces disentangled (i.e., orthogonal to each other)? 3) How do we compute them effectively? This paper provides answers to all these questions. We introduce a robust approach that generates synthetic datasets with c"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Integrating them leads to a reduced LoRA subspace that enables efficient LoRA fine-tuning with improved prediction accuracy for downstream tasks. In particular, we show that such a reduced LoRA subspace, despite being derived entirely from synthetic data, generalizes to real datasets.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The controlled synthetic variations sufficiently capture the statistical structure of real-world 3D variations in texture, geometry, camera motion, and lighting so that subspaces extracted from synthetic data remain useful on real data.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Mining LoRA subspaces from synthetic 3D variation datasets produces approximately orthogonal adapters that combine into an efficient fine-tuning method generalizing from synthetic to real data.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Subspaces extracted from synthetic 3D variations combine into a compact LoRA adapter that raises fine-tuning accuracy on real data","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"5fbbbefd7e88e378adfeed938b05bd85865b81a6504c5bd4b3683365ea4a25c9"},"source":{"id":"2604.10095","kind":"arxiv","version":2},"verdict":{"id":"58f9e89d-f99d-4b2d-85f7-635a70715e34","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T16:24:55.337218Z","strongest_claim":"Integrating them leads to a reduced LoRA subspace that enables efficient LoRA fine-tuning with improved prediction accuracy for downstream tasks. In particular, we show that such a reduced LoRA subspace, despite being derived entirely from synthetic data, generalizes to real datasets.","one_line_summary":"Mining LoRA subspaces from synthetic 3D variation datasets produces approximately orthogonal adapters that combine into an efficient fine-tuning method generalizing from synthetic to real data.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The controlled synthetic variations sufficiently capture the statistical structure of real-world 3D variations in texture, geometry, camera motion, and lighting so that subspaces extracted from synthetic data remain useful on real data.","pith_extraction_headline":"Subspaces extracted from synthetic 3D variations combine into a compact LoRA adapter that raises fine-tuning accuracy on real data"},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.10095/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"},"verdict_id":"58f9e89d-f99d-4b2d-85f7-635a70715e34"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-27T01:05:54Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"VZOgcq7zYdlEa0KAfbW0nIiyeBGMX+Kk8bAiJWwxYlhhogkz6BPBSj938CaFZx0SNvOtwnhJFORwiIQ/Zsj7BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T03:40:15.994718Z"},"content_sha256":"9b5dbdfbad83717e236738e8a3faf4352b8d704b70c5cd603f69356f05306262","schema_version":"1.0","event_id":"sha256:9b5dbdfbad83717e236738e8a3faf4352b8d704b70c5cd603f69356f05306262"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/EKHY7K6GY6BPHII4RY2PZL2LKY/bundle.json","state_url":"https://pith.science/pith/EKHY7K6GY6BPHII4RY2PZL2LKY/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/EKHY7K6GY6BPHII4RY2PZL2LKY/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-08-07T03:40:15Z","links":{"resolver":"https://pith.science/pith/EKHY7K6GY6BPHII4RY2PZL2LKY","bundle":"https://pith.science/pith/EKHY7K6GY6BPHII4RY2PZL2LKY/bundle.json","state":"https://pith.science/pith/EKHY7K6GY6BPHII4RY2PZL2LKY/state.json","well_known_bundle":"https://pith.science/.well-known/pith/EKHY7K6GY6BPHII4RY2PZL2LKY/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:EKHY7K6GY6BPHII4RY2PZL2LKY","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"a64f7a081ae6aefb495c6ccece4d3b026b2afac3cc4f2620a1c107f127e204f3","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-04-11T08:33:13Z","title_canon_sha256":"05fb92cfbed80391deea7759521dcf3128e04b590a0ae7b55794cf2820d9a37d"},"schema_version":"1.0","source":{"id":"2604.10095","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.10095","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"arxiv_version","alias_value":"2604.10095v2","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.10095","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_12","alias_value":"EKHY7K6GY6BP","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_16","alias_value":"EKHY7K6GY6BPHII4","created_at":"2026-05-27T01:05:54Z"},{"alias_kind":"pith_short_8","alias_value":"EKHY7K6G","created_at":"2026-05-27T01:05:54Z"}],"graph_snapshots":[{"event_id":"sha256:9b5dbdfbad83717e236738e8a3faf4352b8d704b70c5cd603f69356f05306262","target":"graph","created_at":"2026-05-27T01:05:54Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Integrating them leads to a reduced LoRA subspace that enables efficient LoRA fine-tuning with improved prediction accuracy for downstream tasks. In particular, we show that such a reduced LoRA subspace, despite being derived entirely from synthetic data, generalizes to real datasets."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The controlled synthetic variations sufficiently capture the statistical structure of real-world 3D variations in texture, geometry, camera motion, and lighting so that subspaces extracted from synthetic data remain useful on real data."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Mining LoRA subspaces from synthetic 3D variation datasets produces approximately orthogonal adapters that combine into an efficient fine-tuning method generalizing from synthetic to real data."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Subspaces extracted from synthetic 3D variations combine into a compact LoRA adapter that raises fine-tuning accuracy on real data"}],"snapshot_sha256":"5fbbbefd7e88e378adfeed938b05bd85865b81a6504c5bd4b3683365ea4a25c9"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2604.10095/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"With the emergence of 3D foundation models, there is growing interest in fine-tuning them for downstream tasks, where LoRA is the dominant fine-tuning paradigm. As 3D datasets exhibit distinct variations in texture, geometry, camera motion, and lighting, there are interesting fundamental questions: 1) Are there LoRA subspaces associated with each type of variation? 2) Are these subspaces disentangled (i.e., orthogonal to each other)? 3) How do we compute them effectively? This paper provides answers to all these questions. We introduce a robust approach that generates synthetic datasets with c","authors_text":"Ahmed Abdelkader, Brandon Y. Feng, Hanwen Jiang, Qixing Huang, Wen-Sheng Chu, Yu Jiang, Zhangyang Wang","cross_cats":[],"headline":"Subspaces extracted from synthetic 3D variations combine into a compact LoRA adapter that raises fine-tuning accuracy on real data","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-04-11T08:33:13Z","title":"Mining Attribute Subspaces for Efficient Fine-tuning of 3D Foundation Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.10095","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-10T16:24:55.337218Z","id":"58f9e89d-f99d-4b2d-85f7-635a70715e34","model_set":{"reader":"grok-4.3"},"one_line_summary":"Mining LoRA subspaces from synthetic 3D variation datasets produces approximately orthogonal adapters that combine into an efficient fine-tuning method generalizing from synthetic to real data.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Subspaces extracted from synthetic 3D variations combine into a compact LoRA adapter that raises fine-tuning accuracy on real data","strongest_claim":"Integrating them leads to a reduced LoRA subspace that enables efficient LoRA fine-tuning with improved prediction accuracy for downstream tasks. In particular, we show that such a reduced LoRA subspace, despite being derived entirely from synthetic data, generalizes to real datasets.","weakest_assumption":"The controlled synthetic variations sufficiently capture the statistical structure of real-world 3D variations in texture, geometry, camera motion, and lighting so that subspaces extracted from synthetic data remain useful on real data."}},"verdict_id":"58f9e89d-f99d-4b2d-85f7-635a70715e34"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:241d7bc7a8a4ff7e40cc79c78ce15d734ba66c1987811240413ecf0a2504d6cc","target":"record","created_at":"2026-05-27T01:05:54Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"a64f7a081ae6aefb495c6ccece4d3b026b2afac3cc4f2620a1c107f127e204f3","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-04-11T08:33:13Z","title_canon_sha256":"05fb92cfbed80391deea7759521dcf3128e04b590a0ae7b55794cf2820d9a37d"},"schema_version":"1.0","source":{"id":"2604.10095","kind":"arxiv","version":2}},"canonical_sha256":"228f8fabc6c782f3a11c8e34fcaf4b5630469a8d59e8678273219d4700351c51","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"228f8fabc6c782f3a11c8e34fcaf4b5630469a8d59e8678273219d4700351c51","first_computed_at":"2026-05-27T01:05:54.207622Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-27T01:05:54.207622Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"MHLajTpRMZuOZM+qna/TIg0/nzuT1cQw4sgwllKkaIUKR65Bekf5VVMHE/MZ9/WVtLAVkHqtZrDQxRn86sYPAg==","signature_status":"signed_v1","signed_at":"2026-05-27T01:05:54.208473Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.10095","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:241d7bc7a8a4ff7e40cc79c78ce15d734ba66c1987811240413ecf0a2504d6cc","sha256:9b5dbdfbad83717e236738e8a3faf4352b8d704b70c5cd603f69356f05306262"],"state_sha256":"179c354a8d39fdceacbcfe14d05f41ed8d55fafaaf6474d09f25a710ea7be738"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"DyX0MMJo2p+JxOLFtTo1sL0/L/74YEca8m+vtic3cUN2jQX+CN2vjZKRAA2UGxFPB1i+bEkyFrwPqZA1QGS5Dw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T03:40:15.999664Z","bundle_sha256":"dc324c27e03a47d9c01c7e7a1fa01d95927507a8fea890f2fb3a812291f65af7"}}