{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:S4WMJOGK34AF5JTC7WSFQWL44B","short_pith_number":"pith:S4WMJOGK","schema_version":"1.0","canonical_sha256":"972cc4b8cadf005ea662fda458597ce05d068e50a87ce1fd1debb4b19719c2b7","source":{"kind":"arxiv","id":"2511.09483","version":3},"attestation_state":"computed","paper":{"title":"CrochetBench: Can Vision-Language Models Move from Describing to Doing in Crochet Domain?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Nitesh V. Chawla, Peiyu Li, Ting Hua, Xiaobao Huang","submitted_at":"2025-11-12T16:52:56Z","abstract_excerpt":"While multimodal large language models can describe visual content, their ability to generate executable procedures remains underexplored. CrochetBench presented in this paper evaluates this shift from describing to doing through fine-grained procedural reasoning in crochet: models must recognize stitches, select structurally appropriate instructions, and generate compilable procedures. We adopt the CrochetPARADE DSL as our intermediate representation, enabling structural validation and functional evaluation via execution. The benchmark covers tasks including stitch classification, instruction"},"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":"2511.09483","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2025-11-12T16:52:56Z","cross_cats_sorted":[],"title_canon_sha256":"c686d78b4741df836e94bd473786969e35accbb9fac0b24b9e3cf9460736f152","abstract_canon_sha256":"34083375c4a4dbf28a4abf727e7811d4e07687d13d3eb92827cdc532cd34181b"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-15T00:21:14.121210Z","signature_b64":"MIPzwq8h4wPmZdTsgpG5oYTfzjV7UlJ3QxxSWjN0EG90/pP7KZ2h7WOmq59S6jQN+G/0lhqSu7gU5TzTMP9SDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"972cc4b8cadf005ea662fda458597ce05d068e50a87ce1fd1debb4b19719c2b7","last_reissued_at":"2026-07-15T00:21:14.120259Z","signature_status":"signed_v1","first_computed_at":"2026-07-15T00:21:14.120259Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"CrochetBench: Can Vision-Language Models Move from Describing to Doing in Crochet Domain?","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Nitesh V. Chawla, Peiyu Li, Ting Hua, Xiaobao Huang","submitted_at":"2025-11-12T16:52:56Z","abstract_excerpt":"While multimodal large language models can describe visual content, their ability to generate executable procedures remains underexplored. CrochetBench presented in this paper evaluates this shift from describing to doing through fine-grained procedural reasoning in crochet: models must recognize stitches, select structurally appropriate instructions, and generate compilable procedures. We adopt the CrochetPARADE DSL as our intermediate representation, enabling structural validation and functional evaluation via execution. The benchmark covers tasks including stitch classification, instruction"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.09483","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/2511.09483/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":"2511.09483","created_at":"2026-07-15T00:21:14.120725+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.09483v3","created_at":"2026-07-15T00:21:14.120725+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.09483","created_at":"2026-07-15T00:21:14.120725+00:00"},{"alias_kind":"pith_short_12","alias_value":"S4WMJOGK34AF","created_at":"2026-07-15T00:21:14.120725+00:00"},{"alias_kind":"pith_short_16","alias_value":"S4WMJOGK34AF5JTC","created_at":"2026-07-15T00:21:14.120725+00:00"},{"alias_kind":"pith_short_8","alias_value":"S4WMJOGK","created_at":"2026-07-15T00:21:14.120725+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B","json":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B.json","graph_json":"https://pith.science/api/pith-number/S4WMJOGK34AF5JTC7WSFQWL44B/graph.json","events_json":"https://pith.science/api/pith-number/S4WMJOGK34AF5JTC7WSFQWL44B/events.json","paper":"https://pith.science/paper/S4WMJOGK"},"agent_actions":{"view_html":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B","download_json":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B.json","view_paper":"https://pith.science/paper/S4WMJOGK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.09483&json=true","fetch_graph":"https://pith.science/api/pith-number/S4WMJOGK34AF5JTC7WSFQWL44B/graph.json","fetch_events":"https://pith.science/api/pith-number/S4WMJOGK34AF5JTC7WSFQWL44B/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B/action/timestamp_anchor","attest_storage":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B/action/storage_attestation","attest_author":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B/action/author_attestation","sign_citation":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B/action/citation_signature","submit_replication":"https://pith.science/pith/S4WMJOGK34AF5JTC7WSFQWL44B/action/replication_record"}},"created_at":"2026-07-15T00:21:14.120725+00:00","updated_at":"2026-07-15T00:21:14.120725+00:00"}