{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2023:EJIVFRXU6P2MSHA7YTIQRPXXRC","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":"c2e103dccd0a0f4cdf6a0868e0ac08d507e2a78eaf6cae4a8431e4e07a4e19f3","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-14T17:57:47Z","title_canon_sha256":"898de40b5c3a0c4369f3ad602ee1292bf94bf954d63805c955271fb60fc783da"},"schema_version":"1.0","source":{"id":"2303.08128","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2303.08128","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"arxiv_version","alias_value":"2303.08128v1","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2303.08128","created_at":"2026-05-17T23:38:13Z"},{"alias_kind":"pith_short_12","alias_value":"EJIVFRXU6P2M","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"EJIVFRXU6P2MSHA7","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"EJIVFRXU","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:062b700cdb153984bb446d7feb17f83f2ef2c73e81beef559b6c9a526afa0ea4","target":"graph","created_at":"2026-05-17T23:38:13Z","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":"This simple approach requires no further training, and achieves state-of-the-art results across various complex visual tasks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That a code-generation model can reliably produce correct, executable compositions of the provided vision modules for arbitrary queries without systematic errors in program logic or API usage."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"ViperGPT generates executable Python code to compose pre-trained vision-and-language modules into programs that answer visual queries, reaching state-of-the-art results with no additional training."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"ViperGPT uses code generation to create Python programs that combine vision models for answering complex visual queries without training."}],"snapshot_sha256":"f15980db7861e4a53f01b42c8d6dfe8448e4e929cb3fe54574e11b2f92be62ff"},"formal_canon":{"evidence_count":1,"snapshot_sha256":"8e12ce63628226e7721f95aaf10ac650aecb66d8c42e4a5cea1dd417ee4dd854"},"paper":{"abstract_excerpt":"Answering visual queries is a complex task that requires both visual processing and reasoning. End-to-end models, the dominant approach for this task, do not explicitly differentiate between the two, limiting interpretability and generalization. Learning modular programs presents a promising alternative, but has proven challenging due to the difficulty of learning both the programs and modules simultaneously. We introduce ViperGPT, a framework that leverages code-generation models to compose vision-and-language models into subroutines to produce a result for any query. ViperGPT utilizes a prov","authors_text":"Carl Vondrick, D\\'idac Sur\\'is, Sachit Menon","cross_cats":[],"headline":"ViperGPT uses code generation to create Python programs that combine vision models for answering complex visual queries without training.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-14T17:57:47Z","title":"ViperGPT: Visual Inference via Python Execution for Reasoning"},"references":{"count":66,"internal_anchors":12,"resolved_work":66,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Flamingo: a visual language model for few-shot learning","work_id":"6c7c3027-f918-40e9-b9e8-c682e15c12a3","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Neural module networks","work_id":"5d0a6c6e-61ca-40d3-9377-b7330e86cf81","year":2016},{"cited_arxiv_id":"1811.12889","doi":"","is_internal_anchor":true,"ref_index":3,"title":"arXiv preprint arXiv:1811.12889 , year=","work_id":"43c4bd82-57c8-4d3e-8e94-66ae1ffab8f3","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"The Consciousness Prior, Dec","work_id":"3d734855-ef5a-4524-afd9-bedb436a7293","year":2019},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Bravo, Sudhanshu Mittal, Simon Ging, and Thomas Brox","work_id":"c6fb2c6a-4031-4584-b576-d3ca0ed8686a","year":2022}],"snapshot_sha256":"3a390387f7b38ac80d23d44eff6c3526995982f9a376079f94e3bc33d09ec62f"},"source":{"id":"2303.08128","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-17T18:09:01.184006Z","id":"58654800-9a63-4a21-a99b-ddaa12a324a7","model_set":{"reader":"grok-4.3"},"one_line_summary":"ViperGPT generates executable Python code to compose pre-trained vision-and-language modules into programs that answer visual queries, reaching state-of-the-art results with no additional training.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"ViperGPT uses code generation to create Python programs that combine vision models for answering complex visual queries without training.","strongest_claim":"This simple approach requires no further training, and achieves state-of-the-art results across various complex visual tasks.","weakest_assumption":"That a code-generation model can reliably produce correct, executable compositions of the provided vision modules for arbitrary queries without systematic errors in program logic or API usage."}},"verdict_id":"58654800-9a63-4a21-a99b-ddaa12a324a7"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:c9b51b989611fe1cea6a236dfa64ef3d4929ac661a5a7f5955a6fea45f1966ae","target":"record","created_at":"2026-05-17T23:38:13Z","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":"c2e103dccd0a0f4cdf6a0868e0ac08d507e2a78eaf6cae4a8431e4e07a4e19f3","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2023-03-14T17:57:47Z","title_canon_sha256":"898de40b5c3a0c4369f3ad602ee1292bf94bf954d63805c955271fb60fc783da"},"schema_version":"1.0","source":{"id":"2303.08128","kind":"arxiv","version":1}},"canonical_sha256":"225152c6f4f3f4c91c1fc4d108bef788a12aea3102ff5ebaa7da712a657626ac","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"225152c6f4f3f4c91c1fc4d108bef788a12aea3102ff5ebaa7da712a657626ac","first_computed_at":"2026-05-17T23:38:13.427145Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:38:13.427145Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"6UPTUj7b+4qmbA5QdZqLcVVAgxhR0tUYGO2LeeC2o8DwvBkHAA0gfsl+XYYTprCSyuo5KvoBw+F43iKP+GgbAQ==","signature_status":"signed_v1","signed_at":"2026-05-17T23:38:13.427971Z","signed_message":"canonical_sha256_bytes"},"source_id":"2303.08128","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:c9b51b989611fe1cea6a236dfa64ef3d4929ac661a5a7f5955a6fea45f1966ae","sha256:062b700cdb153984bb446d7feb17f83f2ef2c73e81beef559b6c9a526afa0ea4"],"state_sha256":"84188bf25389cb8e316755d7fa885eaf562e778bd42015ce0e40f6530c14ae3b"}