{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2024:FPKQ4GTM57XQBF7MR4I7JH3IE6","short_pith_number":"pith:FPKQ4GTM","schema_version":"1.0","canonical_sha256":"2bd50e1a6cefef0097ec8f11f49f6827831c213dd41f2e84942b9cb9fa54fcb6","source":{"kind":"arxiv","id":"2406.09403","version":3},"attestation_state":"computed","paper":{"title":"Visual Sketchpad: Sketching as a Visual Chain of Thought for Multimodal Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Dan Roth, Luke Zettlemoyer, Mari Ostendorf, Noah A Smith, Ranjay Krishna, Weijia Shi, Xingyu Fu, Yushi Hu","submitted_at":"2024-06-13T17:59:31Z","abstract_excerpt":"Humans draw to facilitate reasoning: we draw auxiliary lines when solving geometry problems; we mark and circle when reasoning on maps; we use sketches to amplify our ideas and relieve our limited-capacity working memory. However, such actions are missing in current multimodal language models (LMs). Current chain-of-thought and tool-use paradigms only use text as intermediate reasoning steps. In this work, we introduce Sketchpad, a framework that gives multimodal LMs a visual sketchpad and tools to draw on the sketchpad. The LM conducts planning and reasoning according to the visual artifacts "},"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":"2406.09403","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-06-13T17:59:31Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"367f0b688db05eaaf7daeefad1b50fbe00d4c01824faccf123e9a88a015936fd","abstract_canon_sha256":"f2bca53d46f4340c6c6545c75bf8a59f9e591571a0d699263df09bff6da8eceb"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:33:33.261039Z","signature_b64":"w5I0werXPfYpUAor0fcY2o9ZvfZnAJmMmHxfVUjPWFVulZyrkNS7cCMojWJ2hNuKhIoQDmAWLxji9kcgp0XYBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2bd50e1a6cefef0097ec8f11f49f6827831c213dd41f2e84942b9cb9fa54fcb6","last_reissued_at":"2026-07-05T09:33:33.260493Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:33:33.260493Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Visual Sketchpad: Sketching as a Visual Chain of Thought for Multimodal Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.CV","authors_text":"Dan Roth, Luke Zettlemoyer, Mari Ostendorf, Noah A Smith, Ranjay Krishna, Weijia Shi, Xingyu Fu, Yushi Hu","submitted_at":"2024-06-13T17:59:31Z","abstract_excerpt":"Humans draw to facilitate reasoning: we draw auxiliary lines when solving geometry problems; we mark and circle when reasoning on maps; we use sketches to amplify our ideas and relieve our limited-capacity working memory. However, such actions are missing in current multimodal language models (LMs). Current chain-of-thought and tool-use paradigms only use text as intermediate reasoning steps. In this work, we introduce Sketchpad, a framework that gives multimodal LMs a visual sketchpad and tools to draw on the sketchpad. The LM conducts planning and reasoning according to the visual artifacts "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2406.09403","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/2406.09403/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":"2406.09403","created_at":"2026-07-05T09:33:33.260560+00:00"},{"alias_kind":"arxiv_version","alias_value":"2406.09403v3","created_at":"2026-07-05T09:33:33.260560+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2406.09403","created_at":"2026-07-05T09:33:33.260560+00:00"},{"alias_kind":"pith_short_12","alias_value":"FPKQ4GTM57XQ","created_at":"2026-07-05T09:33:33.260560+00:00"},{"alias_kind":"pith_short_16","alias_value":"FPKQ4GTM57XQBF7M","created_at":"2026-07-05T09:33:33.260560+00:00"},{"alias_kind":"pith_short_8","alias_value":"FPKQ4GTM","created_at":"2026-07-05T09:33:33.260560+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":31,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.25319","citing_title":"V-Zero: Answer-Label-Free On-Policy Distillation with Contrastive Evidence Gating for Fine-Grained Visual Reasoning","ref_index":48,"is_internal_anchor":false},{"citing_arxiv_id":"2606.24233","citing_title":"Latent Visual States for Efficient Multimodal Reasoning","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2606.18974","citing_title":"Visual-OPSD: Cross-Modal On-Policy Self-Distillation for Efficient Unified Multimodal Reasoning","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2606.17888","citing_title":"MathVis-Fine: Aligning Visual Supervision with Necessity via Progressive Dependency-Guided Training for Multimodal Mathematical Reasoning","ref_index":22,"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":96,"is_internal_anchor":false},{"citing_arxiv_id":"2606.05769","citing_title":"Imagine Before You Predict: Interleaved Latent Visual Reasoning for Video Event Prediction","ref_index":37,"is_internal_anchor":false},{"citing_arxiv_id":"2606.03988","citing_title":"Imaginative Perception Tokens Enhance Spatial Reasoning in Multimodal Language Models","ref_index":17,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00384","citing_title":"VESTA: Visual Exploration with Statistical Tool Agents","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2605.27959","citing_title":"ROVER: Routing Object-Centric Visual Evidence for Grounded Multi-Image Reasoning","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28741","citing_title":"Self-Prophetic Decoding to Unlock Visual Search in LVLMs","ref_index":5,"is_internal_anchor":false},{"citing_arxiv_id":"2606.00579","citing_title":"Sandboxed Coding Agents are Competitive Omni-modal Task Solvers","ref_index":2,"is_internal_anchor":false},{"citing_arxiv_id":"2410.04509","citing_title":"ErrorRadar: Benchmarking Complex Mathematical Reasoning of Multimodal Large Language Models Via Error Detection","ref_index":23,"is_internal_anchor":false},{"citing_arxiv_id":"2505.23678","citing_title":"Grounded Reinforcement Learning for Visual Reasoning","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09860","citing_title":"When to Re-Commit: Temporal Abstraction Discovery for Long-Horizon Vision-Language Reasoning","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2605.20743","citing_title":"Draw2Think: Harnessing Geometry Reasoning through Constraint Engine Interaction","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09860","citing_title":"When to Re-Commit: Temporal Abstraction Discovery for Long-Horizon Vision-Language Reasoning","ref_index":22,"is_internal_anchor":false},{"citing_arxiv_id":"2506.06211","citing_title":"PuzzleWorld: A Benchmark for Multimodal, Open-Ended Reasoning in Puzzlehunts","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2503.17352","citing_title":"OpenVLThinker: Complex Vision-Language Reasoning via Iterative SFT-RL Cycles","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2507.01955","citing_title":"How Well Does GPT-4o Understand Vision? Evaluating Multimodal Foundation Models on Standard Computer Vision Tasks","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2501.07542","citing_title":"Imagine while Reasoning in Space: Multimodal Visualization-of-Thought","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2512.12623","citing_title":"Reasoning Within the Mind: Dynamic Multimodal Interleaving in Latent Space","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2512.10941","citing_title":"Mull-Tokens: Modality-Agnostic Latent Thinking","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2503.22020","citing_title":"CoT-VLA: Visual Chain-of-Thought Reasoning for Vision-Language-Action Models","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2509.24251","citing_title":"Latent Visual Reasoning","ref_index":11,"is_internal_anchor":false},{"citing_arxiv_id":"2503.12605","citing_title":"Multimodal Chain-of-Thought Reasoning: A Comprehensive Survey","ref_index":77,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6","json":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6.json","graph_json":"https://pith.science/api/pith-number/FPKQ4GTM57XQBF7MR4I7JH3IE6/graph.json","events_json":"https://pith.science/api/pith-number/FPKQ4GTM57XQBF7MR4I7JH3IE6/events.json","paper":"https://pith.science/paper/FPKQ4GTM"},"agent_actions":{"view_html":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6","download_json":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6.json","view_paper":"https://pith.science/paper/FPKQ4GTM","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2406.09403&json=true","fetch_graph":"https://pith.science/api/pith-number/FPKQ4GTM57XQBF7MR4I7JH3IE6/graph.json","fetch_events":"https://pith.science/api/pith-number/FPKQ4GTM57XQBF7MR4I7JH3IE6/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6/action/storage_attestation","attest_author":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6/action/author_attestation","sign_citation":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6/action/citation_signature","submit_replication":"https://pith.science/pith/FPKQ4GTM57XQBF7MR4I7JH3IE6/action/replication_record"}},"created_at":"2026-07-05T09:33:33.260560+00:00","updated_at":"2026-07-05T09:33:33.260560+00:00"}