{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:FR2DA3TE4WZSP5SDP7MXU4NQKJ","short_pith_number":"pith:FR2DA3TE","schema_version":"1.0","canonical_sha256":"2c74306e64e5b327f6437fd97a71b052647e418e15fb5bd2b59bbf8cd5125461","source":{"kind":"arxiv","id":"2507.10202","version":1},"attestation_state":"computed","paper":{"title":"A Training-Free, Task-Agnostic Framework for Enhancing MLLM Performance on High-Resolution Images","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hanjung Kim, Heechan Choi, Jaeseong Lee, Seonjoo Kim, Yeeun Choi","submitted_at":"2025-07-14T12:14:53Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in vision-language understanding, reasoning, and generation. However, they struggle with tasks requiring fine-grained localization and reasoning in high-resolution images. This constraint stems from the fact that MLLMs are fine-tuned with fixed image resolution to align with the pre-trained image encoder used in MLLM. Consequently, feeding high-resolution images directly into MLLMs leads to poor generalization due to a train-test resolution discrepancy, while downsampling these images-although ensuring consisten"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2507.10202","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","primary_cat":"cs.CV","submitted_at":"2025-07-14T12:14:53Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"ad9857bfb2072d50857836b7997376286898751b8e8d0e7e3905a9a9b1fe467f","abstract_canon_sha256":"0a77b231ed7afe13c7bb42b527f49966e10476e512c9fde5d81eec0dee1bcdcc"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:36:46.885743Z","signature_b64":"59w7lh+HhGcFv+XDTMNDujhFF9ROB2j4Gl+ammf5ckdy4d79etl4W0jpO4frELy3A8QNlV9HZpphyj2aST6EBg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"2c74306e64e5b327f6437fd97a71b052647e418e15fb5bd2b59bbf8cd5125461","last_reissued_at":"2026-07-05T11:36:46.885293Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:36:46.885293Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"A Training-Free, Task-Agnostic Framework for Enhancing MLLM Performance on High-Resolution Images","license":"http://creativecommons.org/licenses/by-nc-sa/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CV","authors_text":"Hanjung Kim, Heechan Choi, Jaeseong Lee, Seonjoo Kim, Yeeun Choi","submitted_at":"2025-07-14T12:14:53Z","abstract_excerpt":"Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities in vision-language understanding, reasoning, and generation. However, they struggle with tasks requiring fine-grained localization and reasoning in high-resolution images. This constraint stems from the fact that MLLMs are fine-tuned with fixed image resolution to align with the pre-trained image encoder used in MLLM. Consequently, feeding high-resolution images directly into MLLMs leads to poor generalization due to a train-test resolution discrepancy, while downsampling these images-although ensuring consisten"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2507.10202","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/2507.10202/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"},"aliases":[{"alias_kind":"arxiv","alias_value":"2507.10202","created_at":"2026-07-05T11:36:46.885365+00:00"},{"alias_kind":"arxiv_version","alias_value":"2507.10202v1","created_at":"2026-07-05T11:36:46.885365+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2507.10202","created_at":"2026-07-05T11:36:46.885365+00:00"},{"alias_kind":"pith_short_12","alias_value":"FR2DA3TE4WZS","created_at":"2026-07-05T11:36:46.885365+00:00"},{"alias_kind":"pith_short_16","alias_value":"FR2DA3TE4WZSP5SD","created_at":"2026-07-05T11:36:46.885365+00:00"},{"alias_kind":"pith_short_8","alias_value":"FR2DA3TE","created_at":"2026-07-05T11:36:46.885365+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.15542","citing_title":"DRS-GUI: Dynamic Region Search for Training-Free GUI Grounding","ref_index":17,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ","json":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ.json","graph_json":"https://pith.science/api/pith-number/FR2DA3TE4WZSP5SDP7MXU4NQKJ/graph.json","events_json":"https://pith.science/api/pith-number/FR2DA3TE4WZSP5SDP7MXU4NQKJ/events.json","paper":"https://pith.science/paper/FR2DA3TE"},"agent_actions":{"view_html":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ","download_json":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ.json","view_paper":"https://pith.science/paper/FR2DA3TE","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2507.10202&json=true","fetch_graph":"https://pith.science/api/pith-number/FR2DA3TE4WZSP5SDP7MXU4NQKJ/graph.json","fetch_events":"https://pith.science/api/pith-number/FR2DA3TE4WZSP5SDP7MXU4NQKJ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ/action/storage_attestation","attest_author":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ/action/author_attestation","sign_citation":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ/action/citation_signature","submit_replication":"https://pith.science/pith/FR2DA3TE4WZSP5SDP7MXU4NQKJ/action/replication_record"}},"created_at":"2026-07-05T11:36:46.885365+00:00","updated_at":"2026-07-05T11:36:46.885365+00:00"}