{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:LHQKVD52BU463ZYD5KRQQ5IWJK","short_pith_number":"pith:LHQKVD52","canonical_record":{"source":{"id":"2512.18951","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-22T01:58:17Z","cross_cats_sorted":[],"title_canon_sha256":"34b5a85b22085aba2953c0d29f5dafd60f5f2471f351242ea4990e5190a9729f","abstract_canon_sha256":"9482d028b1ca83a27d28ab4cbb0092feb22815cab7e40de2524f1225b1e20b78"},"schema_version":"1.0"},"canonical_sha256":"59e0aa8fba0d39ede703eaa30875164abf5bd754e9a0f6c04ac955842078b939","source":{"kind":"arxiv","id":"2512.18951","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.18951","created_at":"2026-05-18T03:09:32Z"},{"alias_kind":"arxiv_version","alias_value":"2512.18951v3","created_at":"2026-05-18T03:09:32Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.18951","created_at":"2026-05-18T03:09:32Z"},{"alias_kind":"pith_short_12","alias_value":"LHQKVD52BU46","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"LHQKVD52BU463ZYD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"LHQKVD52","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:LHQKVD52BU463ZYD5KRQQ5IWJK","target":"record","payload":{"canonical_record":{"source":{"id":"2512.18951","kind":"arxiv","version":3},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-22T01:58:17Z","cross_cats_sorted":[],"title_canon_sha256":"34b5a85b22085aba2953c0d29f5dafd60f5f2471f351242ea4990e5190a9729f","abstract_canon_sha256":"9482d028b1ca83a27d28ab4cbb0092feb22815cab7e40de2524f1225b1e20b78"},"schema_version":"1.0"},"canonical_sha256":"59e0aa8fba0d39ede703eaa30875164abf5bd754e9a0f6c04ac955842078b939","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T03:09:32.485565Z","signature_b64":"ZYuOY2YYUXgwBlAa0j99R1CeM3uwPEn0epg+7Cd1lfMrdTBxCqhGPbw56M8/4512+0dnPe7AHISYe1L7bIP1Cw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"59e0aa8fba0d39ede703eaa30875164abf5bd754e9a0f6c04ac955842078b939","last_reissued_at":"2026-05-18T03:09:32.484890Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T03:09:32.484890Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2512.18951","source_version":3,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:09:32Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"H36sDtFJ8XkzTkmIeX9WZR0KyanQj6x2KboEUaJ8SVgFYirv5/aG9nYMDvwoFawGL+CFR6DAFTEw/bHhm2i5Dg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-26T19:17:03.511687Z"},"content_sha256":"9a93ada046c579ee87ae1df0eabe6f057c9b954a702ffc86a18b2b946fc49a4c","schema_version":"1.0","event_id":"sha256:9a93ada046c579ee87ae1df0eabe6f057c9b954a702ffc86a18b2b946fc49a4c"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:LHQKVD52BU463ZYD5KRQQ5IWJK","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Benchmarking Attribute Discrimination in Infant-Scale Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Infant-trained vision models build strong visual size representations but perform poorly on color discrimination and text-based color grounding.","cross_cats":[],"primary_cat":"cs.LG","authors_text":"Bihan Wen, Patrick Batsell, Satoshi Tsutsui","submitted_at":"2025-12-22T01:58:17Z","abstract_excerpt":"Infants learn not only object categories but also fine-grained visual attributes such as color, size, and texture from limited experience. Prior infant-scale vision--language models have mainly been evaluated on object recognition, leaving open whether they support within-class attribute discrimination. We introduce a controlled benchmark that varies color, size, and texture across 67 everyday object classes using synthetic rendering to decouple attribute values from object identity. We evaluate infant-trained models (CVCL and an infant-trained DINO baseline) against web-scale and ImageNet mod"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"infant-trained models form strong visual representations for size and discriminate texture comparably to other models, but perform poorly on visual color discrimination, and in the text--vision setting they struggle to ground color and show only modest size grounding. In contrast, web-trained vision--language models strongly ground color from text while exhibiting weaker visual size discrimination.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The synthetic rendering procedure successfully decouples attribute values from object identity across the 67 classes so that performance differences reflect attribute discrimination rather than object recognition confounds.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Infant-scale VLMs discriminate size and texture visually but perform poorly on color and struggle to ground attributes in text, while web-scale models excel at color grounding.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Infant-trained vision models build strong visual size representations but perform poorly on color discrimination and text-based color grounding.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"4896a02c040fe9310be69d1dd3ea93e9f81f416fa57a9c7631bee6ac5b90a2d8"},"source":{"id":"2512.18951","kind":"arxiv","version":3},"verdict":{"id":"ac7b41e2-1bfa-4484-9915-d194c6bb33f1","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-16T20:35:04.910210Z","strongest_claim":"infant-trained models form strong visual representations for size and discriminate texture comparably to other models, but perform poorly on visual color discrimination, and in the text--vision setting they struggle to ground color and show only modest size grounding. In contrast, web-trained vision--language models strongly ground color from text while exhibiting weaker visual size discrimination.","one_line_summary":"Infant-scale VLMs discriminate size and texture visually but perform poorly on color and struggle to ground attributes in text, while web-scale models excel at color grounding.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The synthetic rendering procedure successfully decouples attribute values from object identity across the 67 classes so that performance differences reflect attribute discrimination rather than object recognition confounds.","pith_extraction_headline":"Infant-trained vision models build strong visual size representations but perform poorly on color discrimination and text-based color grounding."},"references":{"count":31,"sample":[{"doi":"","year":2025,"title":"Benchmarking Attribute Discrimination in Infant-Scale Vision-Language Models","work_id":"0e236a47-d56e-4908-94d4-6c100a57cf75","ref_index":1,"cited_arxiv_id":"2512.18951","is_internal_anchor":true},{"doi":"","year":null,"title":"BENCHMARK ATTRIBUTE DISCRIMINATION 2.1. Benchmark Design We construct a controlled benchmark that systematically varies three visual attributes—color, size, and texture—across multiple everyday object","work_id":"e752b0e1-0322-461e-a1d9-b01b7525335d","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Same-class, different color (SCDC)","work_id":"dd3160eb-fcbe-4177-814d-6dad7777fc20","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Same-class, different size (SCDS)","work_id":"eaea2137-2d59-4dc5-9639-0854c551a949","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Same-class, different texture (SCDT) In order to evaluate attribute discrimination, each condi- tion is tested under two complementary modes: Prototype tests (image–only):A prototype embedding is comp","work_id":"93903752-e56c-40a8-96dc-fa8182f060e1","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":31,"snapshot_sha256":"d4eea6fcca5737172d08ccd8c84d23ea9e4a94ab66d2bff48cbbcf87a95dd2e0","internal_anchors":2},"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":"ac7b41e2-1bfa-4484-9915-d194c6bb33f1"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T03:09:32Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zs9vEB7Q7FfkrzhPBMVXvCAktwUfE0xJqTlDPCFDCrDTHZMG+FVIhIceE4gLrobFK34nr58DaGi7pT+TwwtLBg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-06-26T19:17:03.512259Z"},"content_sha256":"3e9b1e35c4b09b7d5605987a4632378f2b60b6cf3f79369e62a875c65d4f5888","schema_version":"1.0","event_id":"sha256:3e9b1e35c4b09b7d5605987a4632378f2b60b6cf3f79369e62a875c65d4f5888"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/LHQKVD52BU463ZYD5KRQQ5IWJK/bundle.json","state_url":"https://pith.science/pith/LHQKVD52BU463ZYD5KRQQ5IWJK/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/LHQKVD52BU463ZYD5KRQQ5IWJK/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":"2026-06-26T19:17:03Z","links":{"resolver":"https://pith.science/pith/LHQKVD52BU463ZYD5KRQQ5IWJK","bundle":"https://pith.science/pith/LHQKVD52BU463ZYD5KRQQ5IWJK/bundle.json","state":"https://pith.science/pith/LHQKVD52BU463ZYD5KRQQ5IWJK/state.json","well_known_bundle":"https://pith.science/.well-known/pith/LHQKVD52BU463ZYD5KRQQ5IWJK/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:LHQKVD52BU463ZYD5KRQQ5IWJK","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":"9482d028b1ca83a27d28ab4cbb0092feb22815cab7e40de2524f1225b1e20b78","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-22T01:58:17Z","title_canon_sha256":"34b5a85b22085aba2953c0d29f5dafd60f5f2471f351242ea4990e5190a9729f"},"schema_version":"1.0","source":{"id":"2512.18951","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2512.18951","created_at":"2026-05-18T03:09:32Z"},{"alias_kind":"arxiv_version","alias_value":"2512.18951v3","created_at":"2026-05-18T03:09:32Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2512.18951","created_at":"2026-05-18T03:09:32Z"},{"alias_kind":"pith_short_12","alias_value":"LHQKVD52BU46","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"LHQKVD52BU463ZYD","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"LHQKVD52","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:3e9b1e35c4b09b7d5605987a4632378f2b60b6cf3f79369e62a875c65d4f5888","target":"graph","created_at":"2026-05-18T03:09:32Z","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":"infant-trained models form strong visual representations for size and discriminate texture comparably to other models, but perform poorly on visual color discrimination, and in the text--vision setting they struggle to ground color and show only modest size grounding. In contrast, web-trained vision--language models strongly ground color from text while exhibiting weaker visual size discrimination."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The synthetic rendering procedure successfully decouples attribute values from object identity across the 67 classes so that performance differences reflect attribute discrimination rather than object recognition confounds."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Infant-scale VLMs discriminate size and texture visually but perform poorly on color and struggle to ground attributes in text, while web-scale models excel at color grounding."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Infant-trained vision models build strong visual size representations but perform poorly on color discrimination and text-based color grounding."}],"snapshot_sha256":"4896a02c040fe9310be69d1dd3ea93e9f81f416fa57a9c7631bee6ac5b90a2d8"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"Infants learn not only object categories but also fine-grained visual attributes such as color, size, and texture from limited experience. Prior infant-scale vision--language models have mainly been evaluated on object recognition, leaving open whether they support within-class attribute discrimination. We introduce a controlled benchmark that varies color, size, and texture across 67 everyday object classes using synthetic rendering to decouple attribute values from object identity. We evaluate infant-trained models (CVCL and an infant-trained DINO baseline) against web-scale and ImageNet mod","authors_text":"Bihan Wen, Patrick Batsell, Satoshi Tsutsui","cross_cats":[],"headline":"Infant-trained vision models build strong visual size representations but perform poorly on color discrimination and text-based color grounding.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-22T01:58:17Z","title":"Benchmarking Attribute Discrimination in Infant-Scale Vision-Language Models"},"references":{"count":31,"internal_anchors":2,"resolved_work":31,"sample":[{"cited_arxiv_id":"2512.18951","doi":"","is_internal_anchor":true,"ref_index":1,"title":"Benchmarking Attribute Discrimination in Infant-Scale Vision-Language Models","work_id":"0e236a47-d56e-4908-94d4-6c100a57cf75","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"BENCHMARK ATTRIBUTE DISCRIMINATION 2.1. Benchmark Design We construct a controlled benchmark that systematically varies three visual attributes—color, size, and texture—across multiple everyday object","work_id":"e752b0e1-0322-461e-a1d9-b01b7525335d","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Same-class, different color (SCDC)","work_id":"dd3160eb-fcbe-4177-814d-6dad7777fc20","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Same-class, different size (SCDS)","work_id":"eaea2137-2d59-4dc5-9639-0854c551a949","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Same-class, different texture (SCDT) In order to evaluate attribute discrimination, each condi- tion is tested under two complementary modes: Prototype tests (image–only):A prototype embedding is comp","work_id":"93903752-e56c-40a8-96dc-fa8182f060e1","year":null}],"snapshot_sha256":"d4eea6fcca5737172d08ccd8c84d23ea9e4a94ab66d2bff48cbbcf87a95dd2e0"},"source":{"id":"2512.18951","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-16T20:35:04.910210Z","id":"ac7b41e2-1bfa-4484-9915-d194c6bb33f1","model_set":{"reader":"grok-4.3"},"one_line_summary":"Infant-scale VLMs discriminate size and texture visually but perform poorly on color and struggle to ground attributes in text, while web-scale models excel at color grounding.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Infant-trained vision models build strong visual size representations but perform poorly on color discrimination and text-based color grounding.","strongest_claim":"infant-trained models form strong visual representations for size and discriminate texture comparably to other models, but perform poorly on visual color discrimination, and in the text--vision setting they struggle to ground color and show only modest size grounding. In contrast, web-trained vision--language models strongly ground color from text while exhibiting weaker visual size discrimination.","weakest_assumption":"The synthetic rendering procedure successfully decouples attribute values from object identity across the 67 classes so that performance differences reflect attribute discrimination rather than object recognition confounds."}},"verdict_id":"ac7b41e2-1bfa-4484-9915-d194c6bb33f1"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:9a93ada046c579ee87ae1df0eabe6f057c9b954a702ffc86a18b2b946fc49a4c","target":"record","created_at":"2026-05-18T03:09:32Z","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":"9482d028b1ca83a27d28ab4cbb0092feb22815cab7e40de2524f1225b1e20b78","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.LG","submitted_at":"2025-12-22T01:58:17Z","title_canon_sha256":"34b5a85b22085aba2953c0d29f5dafd60f5f2471f351242ea4990e5190a9729f"},"schema_version":"1.0","source":{"id":"2512.18951","kind":"arxiv","version":3}},"canonical_sha256":"59e0aa8fba0d39ede703eaa30875164abf5bd754e9a0f6c04ac955842078b939","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"59e0aa8fba0d39ede703eaa30875164abf5bd754e9a0f6c04ac955842078b939","first_computed_at":"2026-05-18T03:09:32.484890Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T03:09:32.484890Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"ZYuOY2YYUXgwBlAa0j99R1CeM3uwPEn0epg+7Cd1lfMrdTBxCqhGPbw56M8/4512+0dnPe7AHISYe1L7bIP1Cw==","signature_status":"signed_v1","signed_at":"2026-05-18T03:09:32.485565Z","signed_message":"canonical_sha256_bytes"},"source_id":"2512.18951","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:9a93ada046c579ee87ae1df0eabe6f057c9b954a702ffc86a18b2b946fc49a4c","sha256:3e9b1e35c4b09b7d5605987a4632378f2b60b6cf3f79369e62a875c65d4f5888"],"state_sha256":"f537796ca505028becc865f8c5921bb7bd4006ae4c58a0a203d4b1c8262f1331"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"i5R0nddE9/ZVvYbe78uK4kN6XDBIV2QX2zJWH5Qan4dT9NNuqt3pheS9IipkMvzwePEE0LtFWql6R57stA5DBg==","signed_message":"bundle_sha256_bytes","signed_at":"2026-06-26T19:17:03.514629Z","bundle_sha256":"a8b846ab11298bd0b6dd8a47dc262e8c0a25ccb2d1218df4185ce5d4350189d8"}}