{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:RA6IRTQF4EHGDCTX7KIIXZYAXV","short_pith_number":"pith:RA6IRTQF","schema_version":"1.0","canonical_sha256":"883c88ce05e10e618a77fa908be700bd69f7ecfd89fb7488546e6752c9cefdd2","source":{"kind":"arxiv","id":"2503.19990","version":3},"attestation_state":"computed","paper":{"title":"LEGO-Puzzles: How Good Are MLLMs at Multi-Step Spatial Reasoning?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Haodong Duan, Junyao Gao, Kai Chen, Kaifeng Lyu, Kexian Tang, Wenran Liu, Yanan Sun, Yanhong Zeng, Zhening Xing","submitted_at":"2025-03-25T18:21:07Z","abstract_excerpt":"Multi-step spatial reasoning entails understanding and reasoning about spatial relationships across multiple sequential steps, which is crucial for tackling complex real-world applications, such as robotic manipulation, autonomous navigation, and automated assembly. To assess how well current Multimodal Large Language Models (MLLMs) have acquired this fundamental capability, we introduce LEGO-Puzzles, a scalable benchmark designed to evaluate both spatial understanding and sequential reasoning in MLLMs through LEGO-based tasks. LEGO-Puzzles consists of 1,100 carefully curated visual question-a"},"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":"2503.19990","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-03-25T18:21:07Z","cross_cats_sorted":[],"title_canon_sha256":"08b9fa440c21f5aa9aaafecb094f5138390c5412791bff9bb0db47ef42c0f191","abstract_canon_sha256":"b8d7b71ff4fdd02e8868e3ff3904f458f0860e3b9f129e0e8533a764ce9c15ce"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:24:17.006974Z","signature_b64":"Px4My3rBV7MksD3Nnb67KYrS3STLX+ek6vHhSrSuHc9ohniShdMI/2EtzXnRks0r7AFVtmkxWL+X54ArrPG4CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"883c88ce05e10e618a77fa908be700bd69f7ecfd89fb7488546e6752c9cefdd2","last_reissued_at":"2026-07-05T11:24:17.006458Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:24:17.006458Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"LEGO-Puzzles: How Good Are MLLMs at Multi-Step Spatial Reasoning?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Haodong Duan, Junyao Gao, Kai Chen, Kaifeng Lyu, Kexian Tang, Wenran Liu, Yanan Sun, Yanhong Zeng, Zhening Xing","submitted_at":"2025-03-25T18:21:07Z","abstract_excerpt":"Multi-step spatial reasoning entails understanding and reasoning about spatial relationships across multiple sequential steps, which is crucial for tackling complex real-world applications, such as robotic manipulation, autonomous navigation, and automated assembly. To assess how well current Multimodal Large Language Models (MLLMs) have acquired this fundamental capability, we introduce LEGO-Puzzles, a scalable benchmark designed to evaluate both spatial understanding and sequential reasoning in MLLMs through LEGO-based tasks. LEGO-Puzzles consists of 1,100 carefully curated visual question-a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.19990","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":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2503.19990/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":"2503.19990","created_at":"2026-07-05T11:24:17.006527+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.19990v3","created_at":"2026-07-05T11:24:17.006527+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.19990","created_at":"2026-07-05T11:24:17.006527+00:00"},{"alias_kind":"pith_short_12","alias_value":"RA6IRTQF4EHG","created_at":"2026-07-05T11:24:17.006527+00:00"},{"alias_kind":"pith_short_16","alias_value":"RA6IRTQF4EHGDCTX","created_at":"2026-07-05T11:24:17.006527+00:00"},{"alias_kind":"pith_short_8","alias_value":"RA6IRTQF","created_at":"2026-07-05T11:24:17.006527+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":10,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.20905","citing_title":"Vesta: A Generalist Embodied Reasoning Model","ref_index":107,"is_internal_anchor":false},{"citing_arxiv_id":"2606.11470","citing_title":"The Periodic Table of LLM Reasoning: A Structured Survey of Reasoning Paradigms, Methods, and Failure Modes","ref_index":228,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05702","citing_title":"Seeing Time: Benchmarking Chronological Reasoning and Shortcut Biases in Vision-Language Models","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05445","citing_title":"Brick-Composer: Using MLLMs for Assembly with Diverse Bricks","ref_index":4,"is_internal_anchor":false},{"citing_arxiv_id":"2606.07602","citing_title":"Sample-Efficient Post-Training for LEGO Spatial-Physics Reasoning","ref_index":25,"is_internal_anchor":false},{"citing_arxiv_id":"2605.21625","citing_title":"Flat-Pack Bench: Evaluating Spatio-Temporal Understanding in Large Vision-Language Models through Furniture Assembly","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.22570","citing_title":"VGenST-Bench: A Benchmark for Spatio-Temporal Reasoning via Active Video Synthesis","ref_index":66,"is_internal_anchor":false},{"citing_arxiv_id":"2605.15792","citing_title":"Reversing the Flow: Generation-to-Understanding Synergy in Large Multimodal Models","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18445","citing_title":"What's Holding Back Latent Visual Reasoning?","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2602.11635","citing_title":"Do MLLMs Really Understand Space? A Mathematical Reasoning Evaluation","ref_index":31,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV","json":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV.json","graph_json":"https://pith.science/api/pith-number/RA6IRTQF4EHGDCTX7KIIXZYAXV/graph.json","events_json":"https://pith.science/api/pith-number/RA6IRTQF4EHGDCTX7KIIXZYAXV/events.json","paper":"https://pith.science/paper/RA6IRTQF"},"agent_actions":{"view_html":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV","download_json":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV.json","view_paper":"https://pith.science/paper/RA6IRTQF","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.19990&json=true","fetch_graph":"https://pith.science/api/pith-number/RA6IRTQF4EHGDCTX7KIIXZYAXV/graph.json","fetch_events":"https://pith.science/api/pith-number/RA6IRTQF4EHGDCTX7KIIXZYAXV/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV/action/timestamp_anchor","attest_storage":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV/action/storage_attestation","attest_author":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV/action/author_attestation","sign_citation":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV/action/citation_signature","submit_replication":"https://pith.science/pith/RA6IRTQF4EHGDCTX7KIIXZYAXV/action/replication_record"}},"created_at":"2026-07-05T11:24:17.006527+00:00","updated_at":"2026-07-05T11:24:17.006527+00:00"}