{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:FIGDRADZACIXODEOA27ECC3M5V","short_pith_number":"pith:FIGDRADZ","canonical_record":{"source":{"id":"2605.19792","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-19T12:56:32Z","cross_cats_sorted":[],"title_canon_sha256":"2a5f9862eb5496445685e38a03b2c9caa9f1c14f61d43859d0e2032ddcc8665e","abstract_canon_sha256":"02ad13f9187d0f702711251542f2349c5c92a340048ff9e15f3627e179e08ad2"},"schema_version":"1.0"},"canonical_sha256":"2a0c3880790091770c8e06be410b6ced7290f1dd6c347abb0678b3be193f7b78","source":{"kind":"arxiv","id":"2605.19792","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.19792","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"arxiv_version","alias_value":"2605.19792v1","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19792","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"pith_short_12","alias_value":"FIGDRADZACIX","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"pith_short_16","alias_value":"FIGDRADZACIXODEO","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"pith_short_8","alias_value":"FIGDRADZ","created_at":"2026-05-20T01:06:14Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:FIGDRADZACIXODEOA27ECC3M5V","target":"record","payload":{"canonical_record":{"source":{"id":"2605.19792","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-19T12:56:32Z","cross_cats_sorted":[],"title_canon_sha256":"2a5f9862eb5496445685e38a03b2c9caa9f1c14f61d43859d0e2032ddcc8665e","abstract_canon_sha256":"02ad13f9187d0f702711251542f2349c5c92a340048ff9e15f3627e179e08ad2"},"schema_version":"1.0"},"canonical_sha256":"2a0c3880790091770c8e06be410b6ced7290f1dd6c347abb0678b3be193f7b78","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-20T01:06:14.263873Z","signature_b64":"LBXwGbglTWjX3oGrXmwG661W91t33gWHtXvmwEsoNyHtuxxiaRO8uYWte28sbtfWfKsSutcRkhAXDxR7H7hmBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2a0c3880790091770c8e06be410b6ced7290f1dd6c347abb0678b3be193f7b78","last_reissued_at":"2026-05-20T01:06:14.262886Z","signature_status":"signed_v1","first_computed_at":"2026-05-20T01:06:14.262886Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.19792","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-20T01:06:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"WP+XTrBVzJSWuE5tqlhqvCK/nn72s5BAo8TImt8UVRG4ZA33s6YxX71r7sZ0KRNnX52HODjz/Dwzi0xhCrFsDQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-23T00:28:21.329995Z"},"content_sha256":"e06f5b329e8fe01712c6c6e8aa2629faa138bc8b90fcbf8b3235b73267d8a00d","schema_version":"1.0","event_id":"sha256:e06f5b329e8fe01712c6c6e8aa2629faa138bc8b90fcbf8b3235b73267d8a00d"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:FIGDRADZACIXODEOA27ECC3M5V","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Mechanisms of Object Localization in Vision-Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Gemma Roig, Martina G. Vilas, Timothy Schauml\\\"offel","submitted_at":"2026-05-19T12:56:32Z","abstract_excerpt":"Visually-grounded language models (VLMs) are highly effective in linking visual and textual information, yet they often struggle with basic classification and localization tasks. While classification mechanisms have been studied more extensively, the processes that support object localization remain poorly understood. In this work, we investigate two representative families, LLaVA-1.5 and InternVL-3.5, using a suite of mechanistic interpretability tools, including token ablations, attention knockout, and causal mediation analysis. We find that localization is driven by a containerization mecha"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19792","kind":"arxiv","version":1},"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/2605.19792/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":"2026-05-20T01:06:14Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"61Bxm5w2i2KacZPZNdFc9q/6M2XC75PcshpupvED8AYvNH+wYoFNRRA1duBVGF7rsLqss6S293KuEfP99QcEDw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-23T00:28:21.330572Z"},"content_sha256":"4e416d6003e2cf6acc58c7763f7210240adf4c8f7b9e7f3b532076bd3f741dd3","schema_version":"1.0","event_id":"sha256:4e416d6003e2cf6acc58c7763f7210240adf4c8f7b9e7f3b532076bd3f741dd3"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/FIGDRADZACIXODEOA27ECC3M5V/bundle.json","state_url":"https://pith.science/pith/FIGDRADZACIXODEOA27ECC3M5V/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/FIGDRADZACIXODEOA27ECC3M5V/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":"2026-05-23T00:28:21Z","links":{"resolver":"https://pith.science/pith/FIGDRADZACIXODEOA27ECC3M5V","bundle":"https://pith.science/pith/FIGDRADZACIXODEOA27ECC3M5V/bundle.json","state":"https://pith.science/pith/FIGDRADZACIXODEOA27ECC3M5V/state.json","well_known_bundle":"https://pith.science/.well-known/pith/FIGDRADZACIXODEOA27ECC3M5V/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:FIGDRADZACIXODEOA27ECC3M5V","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":"02ad13f9187d0f702711251542f2349c5c92a340048ff9e15f3627e179e08ad2","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-19T12:56:32Z","title_canon_sha256":"2a5f9862eb5496445685e38a03b2c9caa9f1c14f61d43859d0e2032ddcc8665e"},"schema_version":"1.0","source":{"id":"2605.19792","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.19792","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"arxiv_version","alias_value":"2605.19792v1","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.19792","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"pith_short_12","alias_value":"FIGDRADZACIX","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"pith_short_16","alias_value":"FIGDRADZACIXODEO","created_at":"2026-05-20T01:06:14Z"},{"alias_kind":"pith_short_8","alias_value":"FIGDRADZ","created_at":"2026-05-20T01:06:14Z"}],"graph_snapshots":[{"event_id":"sha256:4e416d6003e2cf6acc58c7763f7210240adf4c8f7b9e7f3b532076bd3f741dd3","target":"graph","created_at":"2026-05-20T01:06:14Z","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/2605.19792/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Visually-grounded language models (VLMs) are highly effective in linking visual and textual information, yet they often struggle with basic classification and localization tasks. While classification mechanisms have been studied more extensively, the processes that support object localization remain poorly understood. In this work, we investigate two representative families, LLaVA-1.5 and InternVL-3.5, using a suite of mechanistic interpretability tools, including token ablations, attention knockout, and causal mediation analysis. We find that localization is driven by a containerization mecha","authors_text":"Gemma Roig, Martina G. Vilas, Timothy Schauml\\\"offel","cross_cats":[],"headline":"","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-19T12:56:32Z","title":"Mechanisms of Object Localization in Vision-Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.19792","kind":"arxiv","version":1},"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:e06f5b329e8fe01712c6c6e8aa2629faa138bc8b90fcbf8b3235b73267d8a00d","target":"record","created_at":"2026-05-20T01:06:14Z","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":"02ad13f9187d0f702711251542f2349c5c92a340048ff9e15f3627e179e08ad2","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-19T12:56:32Z","title_canon_sha256":"2a5f9862eb5496445685e38a03b2c9caa9f1c14f61d43859d0e2032ddcc8665e"},"schema_version":"1.0","source":{"id":"2605.19792","kind":"arxiv","version":1}},"canonical_sha256":"2a0c3880790091770c8e06be410b6ced7290f1dd6c347abb0678b3be193f7b78","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"2a0c3880790091770c8e06be410b6ced7290f1dd6c347abb0678b3be193f7b78","first_computed_at":"2026-05-20T01:06:14.262886Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-20T01:06:14.262886Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"LBXwGbglTWjX3oGrXmwG661W91t33gWHtXvmwEsoNyHtuxxiaRO8uYWte28sbtfWfKsSutcRkhAXDxR7H7hmBA==","signature_status":"signed_v1","signed_at":"2026-05-20T01:06:14.263873Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.19792","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e06f5b329e8fe01712c6c6e8aa2629faa138bc8b90fcbf8b3235b73267d8a00d","sha256:4e416d6003e2cf6acc58c7763f7210240adf4c8f7b9e7f3b532076bd3f741dd3"],"state_sha256":"55d714b411637e7fc0720b30517c2902c0295deac2b9bc9b18fd2d3538f8a2cc"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"uvYJiW56+sM2LAyyqLRiiVljd0WTdC0lx7JDG4EQW99XKvivsJtdF0ET7hqu3ZOwOUGueQ7HEbZFfzwO/Sl/Cw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-23T00:28:21.333150Z","bundle_sha256":"a06535e1f0073840036c0ebd74f538a1e3106e003f207a805beb4aca8421ca23"}}