{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2024:LYDLVNYMZTL376JTRDZ7IDQ7XZ","short_pith_number":"pith:LYDLVNYM","canonical_record":{"source":{"id":"2405.18415","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-05-28T17:57:06Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"dade63f9984b4629c4175dcd6e66e6d3cfb0c548619b029d763d07cc8ac7e154","abstract_canon_sha256":"f9ca9227fdb3a2d3fb16fa812954a4c3209943585972b92ae40929857ff66467"},"schema_version":"1.0"},"canonical_sha256":"5e06bab70cccd7bff93388f3f40e1fbe4355bce3bea78000b0952c1d0a8c7d8d","source":{"kind":"arxiv","id":"2405.18415","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.18415","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"arxiv_version","alias_value":"2405.18415v2","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18415","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"pith_short_12","alias_value":"LYDLVNYMZTL3","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"pith_short_16","alias_value":"LYDLVNYMZTL376JT","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"pith_short_8","alias_value":"LYDLVNYM","created_at":"2026-07-05T09:30:15Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2024:LYDLVNYMZTL376JTRDZ7IDQ7XZ","target":"record","payload":{"canonical_record":{"source":{"id":"2405.18415","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-05-28T17:57:06Z","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"title_canon_sha256":"dade63f9984b4629c4175dcd6e66e6d3cfb0c548619b029d763d07cc8ac7e154","abstract_canon_sha256":"f9ca9227fdb3a2d3fb16fa812954a4c3209943585972b92ae40929857ff66467"},"schema_version":"1.0"},"canonical_sha256":"5e06bab70cccd7bff93388f3f40e1fbe4355bce3bea78000b0952c1d0a8c7d8d","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T09:30:15.449841Z","signature_b64":"WAaECn7p+Ughazs6BW2Fy9W42WW5a6dvmUYF1k16eQvaIY502mxN4DzwiY0bthTw9e7b8n/eooR/T7We1PbIDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"5e06bab70cccd7bff93388f3f40e1fbe4355bce3bea78000b0952c1d0a8c7d8d","last_reissued_at":"2026-07-05T09:30:15.449348Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T09:30:15.449348Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2405.18415","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":"2026-07-05T09:30:15Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5WheROnabtvoG66l5eJ8LYYSQrOvlarZrUVQu2TutmUStbTTYIXOxSdRHOovj7sGeVUh7KqrDlqabVa+Jd97CQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T10:33:34.806424Z"},"content_sha256":"4b40199e619f4e10635751088e83e627257f169b3a65a56cde212cd42f386140","schema_version":"1.0","event_id":"sha256:4b40199e619f4e10635751088e83e627257f169b3a65a56cde212cd42f386140"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2024:LYDLVNYMZTL376JTRDZ7IDQ7XZ","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Why are Visually-Grounded Language Models Bad at Image Classification?","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI","cs.CL","cs.LG"],"primary_cat":"cs.CV","authors_text":"Alyssa Unell, Dhruba Ghosh, Ludwig Schmidt, Serena Yeung-Levy, Xiaohan Wang, Yuchang Su, Yuhui Zhang","submitted_at":"2024-05-28T17:57:06Z","abstract_excerpt":"Image classification is one of the most fundamental capabilities of machine vision intelligence. In this work, we revisit the image classification task using visually-grounded language models (VLMs) such as GPT-4V and LLaVA. We find that existing proprietary and public VLMs, despite often using CLIP as a vision encoder and having many more parameters, significantly underperform CLIP on standard image classification benchmarks like ImageNet. To understand the reason, we explore several hypotheses concerning the inference algorithms, training objectives, and data processing in VLMs. Our analysis"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18415","kind":"arxiv","version":2},"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/2405.18415/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":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T09:30:15Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"O6uDJ6jrDkzwXTkgpdiwOPh3zYAH86x3aa9FEi9gRXEKx1ezZWGDtLJhjE0mYexBE1k9RnpXQh/WTeoNQSQuCA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-09T10:33:34.806944Z"},"content_sha256":"0cd64343372a1db676540b91a8fe6a6b51ca8c3c8bfe5ff990420d865e5bb45f","schema_version":"1.0","event_id":"sha256:0cd64343372a1db676540b91a8fe6a6b51ca8c3c8bfe5ff990420d865e5bb45f"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/LYDLVNYMZTL376JTRDZ7IDQ7XZ/bundle.json","state_url":"https://pith.science/pith/LYDLVNYMZTL376JTRDZ7IDQ7XZ/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/LYDLVNYMZTL376JTRDZ7IDQ7XZ/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":"2026-08-09T10:33:34Z","links":{"resolver":"https://pith.science/pith/LYDLVNYMZTL376JTRDZ7IDQ7XZ","bundle":"https://pith.science/pith/LYDLVNYMZTL376JTRDZ7IDQ7XZ/bundle.json","state":"https://pith.science/pith/LYDLVNYMZTL376JTRDZ7IDQ7XZ/state.json","well_known_bundle":"https://pith.science/.well-known/pith/LYDLVNYMZTL376JTRDZ7IDQ7XZ/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2024:LYDLVNYMZTL376JTRDZ7IDQ7XZ","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":"f9ca9227fdb3a2d3fb16fa812954a4c3209943585972b92ae40929857ff66467","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-05-28T17:57:06Z","title_canon_sha256":"dade63f9984b4629c4175dcd6e66e6d3cfb0c548619b029d763d07cc8ac7e154"},"schema_version":"1.0","source":{"id":"2405.18415","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2405.18415","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"arxiv_version","alias_value":"2405.18415v2","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2405.18415","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"pith_short_12","alias_value":"LYDLVNYMZTL3","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"pith_short_16","alias_value":"LYDLVNYMZTL376JT","created_at":"2026-07-05T09:30:15Z"},{"alias_kind":"pith_short_8","alias_value":"LYDLVNYM","created_at":"2026-07-05T09:30:15Z"}],"graph_snapshots":[{"event_id":"sha256:0cd64343372a1db676540b91a8fe6a6b51ca8c3c8bfe5ff990420d865e5bb45f","target":"graph","created_at":"2026-07-05T09:30:15Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2405.18415/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Image classification is one of the most fundamental capabilities of machine vision intelligence. In this work, we revisit the image classification task using visually-grounded language models (VLMs) such as GPT-4V and LLaVA. We find that existing proprietary and public VLMs, despite often using CLIP as a vision encoder and having many more parameters, significantly underperform CLIP on standard image classification benchmarks like ImageNet. To understand the reason, we explore several hypotheses concerning the inference algorithms, training objectives, and data processing in VLMs. Our analysis","authors_text":"Alyssa Unell, Dhruba Ghosh, Ludwig Schmidt, Serena Yeung-Levy, Xiaohan Wang, Yuchang Su, Yuhui Zhang","cross_cats":["cs.AI","cs.CL","cs.LG"],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-05-28T17:57:06Z","title":"Why are Visually-Grounded Language Models Bad at Image Classification?"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2405.18415","kind":"arxiv","version":2},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:4b40199e619f4e10635751088e83e627257f169b3a65a56cde212cd42f386140","target":"record","created_at":"2026-07-05T09:30:15Z","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":"f9ca9227fdb3a2d3fb16fa812954a4c3209943585972b92ae40929857ff66467","cross_cats_sorted":["cs.AI","cs.CL","cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2024-05-28T17:57:06Z","title_canon_sha256":"dade63f9984b4629c4175dcd6e66e6d3cfb0c548619b029d763d07cc8ac7e154"},"schema_version":"1.0","source":{"id":"2405.18415","kind":"arxiv","version":2}},"canonical_sha256":"5e06bab70cccd7bff93388f3f40e1fbe4355bce3bea78000b0952c1d0a8c7d8d","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"5e06bab70cccd7bff93388f3f40e1fbe4355bce3bea78000b0952c1d0a8c7d8d","first_computed_at":"2026-07-05T09:30:15.449348Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T09:30:15.449348Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"WAaECn7p+Ughazs6BW2Fy9W42WW5a6dvmUYF1k16eQvaIY502mxN4DzwiY0bthTw9e7b8n/eooR/T7We1PbIDQ==","signature_status":"signed_v1","signed_at":"2026-07-05T09:30:15.449841Z","signed_message":"canonical_sha256_bytes"},"source_id":"2405.18415","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:4b40199e619f4e10635751088e83e627257f169b3a65a56cde212cd42f386140","sha256:0cd64343372a1db676540b91a8fe6a6b51ca8c3c8bfe5ff990420d865e5bb45f"],"state_sha256":"4efc3470f76d1958f202723d7deb5d2a488dfe33753d8ef646b61fff35840ea8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"UtJ4XWznivg9X/jbgzPoBhruTyQzYblSTT072d0/H6qUz6S0Ei8cAd4c9LJ71/0cwykjd/LDNt6E9TMvYZ89DA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-09T10:33:34.810228Z","bundle_sha256":"848969ea5c6a6875f3e67b7bf58f2e4564189f4d35680e1887852d9ccc189da4"}}