{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:SJIMQTE5P6ZIPBRFNFHFOLT56Y","short_pith_number":"pith:SJIMQTE5","canonical_record":{"source":{"id":"2604.18347","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-04-20T14:42:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b52b984bb1460a38f52e7072952fd072158ba0c7b28d1c57ce19d81a88610a7d","abstract_canon_sha256":"fe0948cad938ee6330c6f1745800089e0a96365e70df7c12e7a9dd865e50a475"},"schema_version":"1.0"},"canonical_sha256":"9250c84c9d7fb2878625694e572e7df62bc4448a8102a69d9c7060f43016954e","source":{"kind":"arxiv","id":"2604.18347","version":2},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.18347","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"arxiv_version","alias_value":"2604.18347v2","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.18347","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"pith_short_12","alias_value":"SJIMQTE5P6ZI","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"pith_short_16","alias_value":"SJIMQTE5P6ZIPBRF","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"pith_short_8","alias_value":"SJIMQTE5","created_at":"2026-06-09T01:04:43Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:SJIMQTE5P6ZIPBRFNFHFOLT56Y","target":"record","payload":{"canonical_record":{"source":{"id":"2604.18347","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-04-20T14:42:47Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"b52b984bb1460a38f52e7072952fd072158ba0c7b28d1c57ce19d81a88610a7d","abstract_canon_sha256":"fe0948cad938ee6330c6f1745800089e0a96365e70df7c12e7a9dd865e50a475"},"schema_version":"1.0"},"canonical_sha256":"9250c84c9d7fb2878625694e572e7df62bc4448a8102a69d9c7060f43016954e","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-09T01:04:43.094833Z","signature_b64":"7z1JrSMuAKwHCrtpLAd3baQJ3EfwCoMRiu0VSxxo5ZDQca6GnoZE+EbIU8XuNnzB/NH048smgTaGhLhNxOqrDg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"9250c84c9d7fb2878625694e572e7df62bc4448a8102a69d9c7060f43016954e","last_reissued_at":"2026-06-09T01:04:43.094412Z","signature_status":"signed_v1","first_computed_at":"2026-06-09T01:04:43.094412Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2604.18347","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":"2026-06-09T01:04:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"v2Zs8zw3mkgF6YqSSTPRqpisskqEzJHg3sreGuQfFsWFo2d6R+u57wegjp+Ukl19HvG/J1SxMviZq0QAtz4+Cw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-31T20:03:53.530520Z"},"content_sha256":"813b20319ac46080ea859d0067d76305b73b5f318005d4b2862b2b3ea7818b5a","schema_version":"1.0","event_id":"sha256:813b20319ac46080ea859d0067d76305b73b5f318005d4b2862b2b3ea7818b5a"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:SJIMQTE5P6ZIPBRFNFHFOLT56Y","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Multilingual Training and Evaluation Resources for Vision-Language Models","license":"http://creativecommons.org/licenses/by-sa/4.0/","headline":"Training vision-language models on multilingual multimodal examples improves non-English benchmark performance with positive transfer to English.","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Andrea Zugarini, Daniela Baiamonte, Elena Fano, Leonardo Rigutini, Matteo Gabburo, Stefano Simonazzi","submitted_at":"2026-04-20T14:42:47Z","abstract_excerpt":"Vision Language Models (VLMs) achieved rapid progress in the recent years. However, despite their growth, VLMs development is heavily grounded on English, leading to two main limitations: (i) the lack of multilingual and multimodal datasets for training, and (ii) the scarcity of comprehensive evaluation benchmarks across languages. In this work, we address these gaps by introducing a new comprehensive suite of resources for VLMs training and evaluation spanning five European languages (English, French, German, Italian, and Spanish). We adopt a regeneration-translation paradigm that produces hi"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Experiments, comprising 3 different models show that using multilingual, multimodal examples for training VLMs aids is consistently beneficial on non-English benchmarks, with positive transfer to English as well.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The regeneration-translation process using permissively licensed models produces data of sufficiently high quality and semantic fidelity to serve as effective training and evaluation material equivalent to native-language resources.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Releases regenerated multilingual training data and translated benchmarks for VLMs in five languages and demonstrates consistent benefits from multilingual training over English-only baselines.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Training vision-language models on multilingual multimodal examples improves non-English benchmark performance with positive transfer to English.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"f8545bf7538b3dfef2ed3c6b8937ae2e7bef18dbf785af9a5e8e8422e2369171"},"source":{"id":"2604.18347","kind":"arxiv","version":2},"verdict":{"id":"449fa840-b0d5-4726-9747-4df995d27188","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-10T05:06:50.077568Z","strongest_claim":"Experiments, comprising 3 different models show that using multilingual, multimodal examples for training VLMs aids is consistently beneficial on non-English benchmarks, with positive transfer to English as well.","one_line_summary":"Releases regenerated multilingual training data and translated benchmarks for VLMs in five languages and demonstrates consistent benefits from multilingual training over English-only baselines.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The regeneration-translation process using permissively licensed models produces data of sufficiently high quality and semantic fidelity to serve as effective training and evaluation material equivalent to native-language resources.","pith_extraction_headline":"Training vision-language models on multilingual multimodal examples improves non-English benchmark performance with positive transfer to English."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2604.18347/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_compliance","ran_at":"2026-05-20T04:09:25.881192Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"00759bb585dc4fff47c03c91e657f6c46f7ef53c5705a94b2402398bf1db9a20"},"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":"449fa840-b0d5-4726-9747-4df995d27188"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-06-09T01:04:43Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sn/N54GVTt9ET6vBAyoDnpoah0goTrSPgLSL+WbXM1xGNH7a08Na2bG/lNDkeB50qukPAC9nJfpyc/8UpfxLAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-07-31T20:03:53.531599Z"},"content_sha256":"83bd9aad0ead09853494139b8f661a1f6001d36909bd2eb920b91849ba7097c7","schema_version":"1.0","event_id":"sha256:83bd9aad0ead09853494139b8f661a1f6001d36909bd2eb920b91849ba7097c7"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/SJIMQTE5P6ZIPBRFNFHFOLT56Y/bundle.json","state_url":"https://pith.science/pith/SJIMQTE5P6ZIPBRFNFHFOLT56Y/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/SJIMQTE5P6ZIPBRFNFHFOLT56Y/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":"2026-07-31T20:03:53Z","links":{"resolver":"https://pith.science/pith/SJIMQTE5P6ZIPBRFNFHFOLT56Y","bundle":"https://pith.science/pith/SJIMQTE5P6ZIPBRFNFHFOLT56Y/bundle.json","state":"https://pith.science/pith/SJIMQTE5P6ZIPBRFNFHFOLT56Y/state.json","well_known_bundle":"https://pith.science/.well-known/pith/SJIMQTE5P6ZIPBRFNFHFOLT56Y/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:SJIMQTE5P6ZIPBRFNFHFOLT56Y","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":"fe0948cad938ee6330c6f1745800089e0a96365e70df7c12e7a9dd865e50a475","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-04-20T14:42:47Z","title_canon_sha256":"b52b984bb1460a38f52e7072952fd072158ba0c7b28d1c57ce19d81a88610a7d"},"schema_version":"1.0","source":{"id":"2604.18347","kind":"arxiv","version":2}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2604.18347","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"arxiv_version","alias_value":"2604.18347v2","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2604.18347","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"pith_short_12","alias_value":"SJIMQTE5P6ZI","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"pith_short_16","alias_value":"SJIMQTE5P6ZIPBRF","created_at":"2026-06-09T01:04:43Z"},{"alias_kind":"pith_short_8","alias_value":"SJIMQTE5","created_at":"2026-06-09T01:04:43Z"}],"graph_snapshots":[{"event_id":"sha256:83bd9aad0ead09853494139b8f661a1f6001d36909bd2eb920b91849ba7097c7","target":"graph","created_at":"2026-06-09T01:04:43Z","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":"Experiments, comprising 3 different models show that using multilingual, multimodal examples for training VLMs aids is consistently beneficial on non-English benchmarks, with positive transfer to English as well."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The regeneration-translation process using permissively licensed models produces data of sufficiently high quality and semantic fidelity to serve as effective training and evaluation material equivalent to native-language resources."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"Releases regenerated multilingual training data and translated benchmarks for VLMs in five languages and demonstrates consistent benefits from multilingual training over English-only baselines."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Training vision-language models on multilingual multimodal examples improves non-English benchmark performance with positive transfer to English."}],"snapshot_sha256":"f8545bf7538b3dfef2ed3c6b8937ae2e7bef18dbf785af9a5e8e8422e2369171"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-20T04:09:25.881192Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2604.18347/integrity.json","findings":[],"snapshot_sha256":"00759bb585dc4fff47c03c91e657f6c46f7ef53c5705a94b2402398bf1db9a20","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Vision Language Models (VLMs) achieved rapid progress in the recent years. However, despite their growth, VLMs development is heavily grounded on English, leading to two main limitations: (i) the lack of multilingual and multimodal datasets for training, and (ii) the scarcity of comprehensive evaluation benchmarks across languages. In this work, we address these gaps by introducing a new comprehensive suite of resources for VLMs training and evaluation spanning five European languages (English, French, German, Italian, and Spanish). We adopt a regeneration-translation paradigm that produces hi","authors_text":"Andrea Zugarini, Daniela Baiamonte, Elena Fano, Leonardo Rigutini, Matteo Gabburo, Stefano Simonazzi","cross_cats":["cs.AI"],"headline":"Training vision-language models on multilingual multimodal examples improves non-English benchmark performance with positive transfer to English.","license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-04-20T14:42:47Z","title":"Multilingual Training and Evaluation Resources for Vision-Language Models"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2604.18347","kind":"arxiv","version":2},"verdict":{"created_at":"2026-05-10T05:06:50.077568Z","id":"449fa840-b0d5-4726-9747-4df995d27188","model_set":{"reader":"grok-4.3"},"one_line_summary":"Releases regenerated multilingual training data and translated benchmarks for VLMs in five languages and demonstrates consistent benefits from multilingual training over English-only baselines.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Training vision-language models on multilingual multimodal examples improves non-English benchmark performance with positive transfer to English.","strongest_claim":"Experiments, comprising 3 different models show that using multilingual, multimodal examples for training VLMs aids is consistently beneficial on non-English benchmarks, with positive transfer to English as well.","weakest_assumption":"The regeneration-translation process using permissively licensed models produces data of sufficiently high quality and semantic fidelity to serve as effective training and evaluation material equivalent to native-language resources."}},"verdict_id":"449fa840-b0d5-4726-9747-4df995d27188"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:813b20319ac46080ea859d0067d76305b73b5f318005d4b2862b2b3ea7818b5a","target":"record","created_at":"2026-06-09T01:04:43Z","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":"fe0948cad938ee6330c6f1745800089e0a96365e70df7c12e7a9dd865e50a475","cross_cats_sorted":["cs.AI"],"license":"http://creativecommons.org/licenses/by-sa/4.0/","primary_cat":"cs.CL","submitted_at":"2026-04-20T14:42:47Z","title_canon_sha256":"b52b984bb1460a38f52e7072952fd072158ba0c7b28d1c57ce19d81a88610a7d"},"schema_version":"1.0","source":{"id":"2604.18347","kind":"arxiv","version":2}},"canonical_sha256":"9250c84c9d7fb2878625694e572e7df62bc4448a8102a69d9c7060f43016954e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"9250c84c9d7fb2878625694e572e7df62bc4448a8102a69d9c7060f43016954e","first_computed_at":"2026-06-09T01:04:43.094412Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-06-09T01:04:43.094412Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"7z1JrSMuAKwHCrtpLAd3baQJ3EfwCoMRiu0VSxxo5ZDQca6GnoZE+EbIU8XuNnzB/NH048smgTaGhLhNxOqrDg==","signature_status":"signed_v1","signed_at":"2026-06-09T01:04:43.094833Z","signed_message":"canonical_sha256_bytes"},"source_id":"2604.18347","source_kind":"arxiv","source_version":2}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:813b20319ac46080ea859d0067d76305b73b5f318005d4b2862b2b3ea7818b5a","sha256:83bd9aad0ead09853494139b8f661a1f6001d36909bd2eb920b91849ba7097c7"],"state_sha256":"cc64497088f4cdd52f593c20f3ddd33e8851ecddbfd4de9348b51052ef0a03e8"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"90uGRDxfttVth5BNVJjKAfmKkHYWZk7ZlCqoxKJA6itLAc5YRb3F52yRtYjyovbAhrmmwM/v/L4SAzLP3+9kCw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-07-31T20:03:53.538469Z","bundle_sha256":"ec8b4817cf8cc30a710f4f4fa85ddf89e40864f80f3516478229061e4a907a90"}}