{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:G4CSUWHDN6JZNJ3WCMIB6XLS2I","short_pith_number":"pith:G4CSUWHD","canonical_record":{"source":{"id":"2605.00392","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-01T04:30:16Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"a92fc25361f723038ba98b4e0329bea793aea9d31f4909d71ddfebcc635c0a0c","abstract_canon_sha256":"098170005d634e2acc0930416f394483e1c5e24fbe27629fdd2c002129c26e5e"},"schema_version":"1.0"},"canonical_sha256":"37052a58e36f9396a77613101f5d72d22f0c36fb3f6651797da4d843074d4778","source":{"kind":"arxiv","id":"2605.00392","version":3},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.00392","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"arxiv_version","alias_value":"2605.00392v3","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.00392","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"pith_short_12","alias_value":"G4CSUWHDN6JZ","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"pith_short_16","alias_value":"G4CSUWHDN6JZNJ3W","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"pith_short_8","alias_value":"G4CSUWHD","created_at":"2026-05-22T01:04:03Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:G4CSUWHDN6JZNJ3WCMIB6XLS2I","target":"record","payload":{"canonical_record":{"source":{"id":"2605.00392","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-01T04:30:16Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"a92fc25361f723038ba98b4e0329bea793aea9d31f4909d71ddfebcc635c0a0c","abstract_canon_sha256":"098170005d634e2acc0930416f394483e1c5e24fbe27629fdd2c002129c26e5e"},"schema_version":"1.0"},"canonical_sha256":"37052a58e36f9396a77613101f5d72d22f0c36fb3f6651797da4d843074d4778","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-22T01:04:03.815135Z","signature_b64":"N3UGgl89c0Thbew9b2B609ODApF9/rGvUd+7jt7nBiRegObxh81eH8eEIQMkiQWQ9o5IQQ/qi/BlUwSKh67qDQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"37052a58e36f9396a77613101f5d72d22f0c36fb3f6651797da4d843074d4778","last_reissued_at":"2026-05-22T01:04:03.814269Z","signature_status":"signed_v1","first_computed_at":"2026-05-22T01:04:03.814269Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.00392","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":"2026-05-22T01:04:03Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"4p3UnDJfvrTXULIRSVfc1yCAh0ghn8haVC4yXTSBCT4w4aONVzZiAxQ/P/FBCRaElaXGlU/I7M34kJwG+cZ4Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T10:22:23.885713Z"},"content_sha256":"e0b420d86661fd78a65dc2eca3a647eb49e4c9b90a2a9f9ecf5b3bb98a1ba6b5","schema_version":"1.0","event_id":"sha256:e0b420d86661fd78a65dc2eca3a647eb49e4c9b90a2a9f9ecf5b3bb98a1ba6b5"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:G4CSUWHDN6JZNJ3WCMIB6XLS2I","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"RTPrune: Reading-Twice Inspired Token Pruning for Efficient DeepSeek-OCR Inference","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"RTPrune applies a two-stage pruning process to DeepSeek-OCR that first keeps high-norm visual tokens and then merges the rest with optimal transport to cut inference time while preserving OCR accuracy.","cross_cats":["cs.LG"],"primary_cat":"cs.CV","authors_text":"Ben Wan, Jia Wang, Tongxuan Liu, Weizhe Huang, Yan Feng, Yuting Zeng, Zihan Tang","submitted_at":"2026-05-01T04:30:16Z","abstract_excerpt":"DeepSeek-OCR leverages visual-text compression to reduce long-text processing costs and accelerate inference, yet visual tokens remain prone to redundant textual and structural information. Moreover, current token pruning methods for conventional vision-language models (VLMs) fail to preserve textual fidelity due to improper compression mechanisms. By analyzing the decoding process of DeepSeek-OCR, we find that a distinct two-stage reading trajectory: the model initially prioritizes the majority of high-norm tokens, then subsequently redistributes its attention to the remaining ones. Motivated"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Extensive experiments demonstrate state-of-the-art performance, as evidenced by 99.47% accuracy and 1.23× faster prefill on OmniDocBench, achieved with 84.25% token retention when applied to DeepSeek-OCR-Large.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The observed two-stage reading trajectory in DeepSeek-OCR is stable and general enough that pruning high-norm tokens first followed by optimal-transport merging of the rest will preserve textual fidelity across OCR tasks.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RTPrune delivers 99.47% accuracy and 1.23x faster prefill on OmniDocBench for DeepSeek-OCR-Large by retaining only 84.25% of tokens through a reading-twice inspired two-stage pruning process.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"RTPrune applies a two-stage pruning process to DeepSeek-OCR that first keeps high-norm visual tokens and then merges the rest with optimal transport to cut inference time while preserving OCR accuracy.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"0d9d8ba1dc9964d0b82726a30d44176d8ff830b52192238e23995f1187f20bbf"},"source":{"id":"2605.00392","kind":"arxiv","version":3},"verdict":{"id":"9ad105ae-5aa3-48ec-8548-c6580e248ac8","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-12T03:07:12.746382Z","strongest_claim":"Extensive experiments demonstrate state-of-the-art performance, as evidenced by 99.47% accuracy and 1.23× faster prefill on OmniDocBench, achieved with 84.25% token retention when applied to DeepSeek-OCR-Large.","one_line_summary":"RTPrune delivers 99.47% accuracy and 1.23x faster prefill on OmniDocBench for DeepSeek-OCR-Large by retaining only 84.25% of tokens through a reading-twice inspired two-stage pruning process.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The observed two-stage reading trajectory in DeepSeek-OCR is stable and general enough that pruning high-norm tokens first followed by optimal-transport merging of the rest will preserve textual fidelity across OCR tasks.","pith_extraction_headline":"RTPrune applies a two-stage pruning process to DeepSeek-OCR that first keeps high-norm visual tokens and then merges the rest with optimal transport to cut inference time while preserving OCR accuracy."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.00392/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"ai_meta_artifact","ran_at":"2026-05-20T19:43:24.710508Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T18:12:25.211188Z","status":"completed","version":"1.0.0","findings_count":0}],"snapshot_sha256":"dbc464ed383eb8a369f4a6f19ae2bc067297355c3aa12322dc69e210380a06c9"},"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":"9ad105ae-5aa3-48ec-8548-c6580e248ac8"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-22T01:04:03Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"zSnd4zF8jyOYomtibjGyPLx+HUw++t49GH3Ky1yp/TsHDqST6VrWchtgGDQUOyTGu+BSc8bniRgAauOHPiQZAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-25T10:22:23.886676Z"},"content_sha256":"3b11ff6995537b58074d3f5ba056c113de3fc654c68ca3d7d6ad12d4da7ade2b","schema_version":"1.0","event_id":"sha256:3b11ff6995537b58074d3f5ba056c113de3fc654c68ca3d7d6ad12d4da7ade2b"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/G4CSUWHDN6JZNJ3WCMIB6XLS2I/bundle.json","state_url":"https://pith.science/pith/G4CSUWHDN6JZNJ3WCMIB6XLS2I/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/G4CSUWHDN6JZNJ3WCMIB6XLS2I/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":"2026-05-25T10:22:23Z","links":{"resolver":"https://pith.science/pith/G4CSUWHDN6JZNJ3WCMIB6XLS2I","bundle":"https://pith.science/pith/G4CSUWHDN6JZNJ3WCMIB6XLS2I/bundle.json","state":"https://pith.science/pith/G4CSUWHDN6JZNJ3WCMIB6XLS2I/state.json","well_known_bundle":"https://pith.science/.well-known/pith/G4CSUWHDN6JZNJ3WCMIB6XLS2I/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:G4CSUWHDN6JZNJ3WCMIB6XLS2I","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":"098170005d634e2acc0930416f394483e1c5e24fbe27629fdd2c002129c26e5e","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-01T04:30:16Z","title_canon_sha256":"a92fc25361f723038ba98b4e0329bea793aea9d31f4909d71ddfebcc635c0a0c"},"schema_version":"1.0","source":{"id":"2605.00392","kind":"arxiv","version":3}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.00392","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"arxiv_version","alias_value":"2605.00392v3","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.00392","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"pith_short_12","alias_value":"G4CSUWHDN6JZ","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"pith_short_16","alias_value":"G4CSUWHDN6JZNJ3W","created_at":"2026-05-22T01:04:03Z"},{"alias_kind":"pith_short_8","alias_value":"G4CSUWHD","created_at":"2026-05-22T01:04:03Z"}],"graph_snapshots":[{"event_id":"sha256:3b11ff6995537b58074d3f5ba056c113de3fc654c68ca3d7d6ad12d4da7ade2b","target":"graph","created_at":"2026-05-22T01:04:03Z","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":"Extensive experiments demonstrate state-of-the-art performance, as evidenced by 99.47% accuracy and 1.23× faster prefill on OmniDocBench, achieved with 84.25% token retention when applied to DeepSeek-OCR-Large."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The observed two-stage reading trajectory in DeepSeek-OCR is stable and general enough that pruning high-norm tokens first followed by optimal-transport merging of the rest will preserve textual fidelity across OCR tasks."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"RTPrune delivers 99.47% accuracy and 1.23x faster prefill on OmniDocBench for DeepSeek-OCR-Large by retaining only 84.25% of tokens through a reading-twice inspired two-stage pruning process."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"RTPrune applies a two-stage pruning process to DeepSeek-OCR that first keeps high-norm visual tokens and then merges the rest with optimal transport to cut inference time while preserving OCR accuracy."}],"snapshot_sha256":"0d9d8ba1dc9964d0b82726a30d44176d8ff830b52192238e23995f1187f20bbf"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[{"findings_count":0,"name":"ai_meta_artifact","ran_at":"2026-05-20T19:43:24.710508Z","status":"completed","version":"1.0.0"},{"findings_count":0,"name":"doi_compliance","ran_at":"2026-05-19T18:12:25.211188Z","status":"completed","version":"1.0.0"}],"endpoint":"/pith/2605.00392/integrity.json","findings":[],"snapshot_sha256":"dbc464ed383eb8a369f4a6f19ae2bc067297355c3aa12322dc69e210380a06c9","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"DeepSeek-OCR leverages visual-text compression to reduce long-text processing costs and accelerate inference, yet visual tokens remain prone to redundant textual and structural information. Moreover, current token pruning methods for conventional vision-language models (VLMs) fail to preserve textual fidelity due to improper compression mechanisms. By analyzing the decoding process of DeepSeek-OCR, we find that a distinct two-stage reading trajectory: the model initially prioritizes the majority of high-norm tokens, then subsequently redistributes its attention to the remaining ones. Motivated","authors_text":"Ben Wan, Jia Wang, Tongxuan Liu, Weizhe Huang, Yan Feng, Yuting Zeng, Zihan Tang","cross_cats":["cs.LG"],"headline":"RTPrune applies a two-stage pruning process to DeepSeek-OCR that first keeps high-norm visual tokens and then merges the rest with optimal transport to cut inference time while preserving OCR accuracy.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-01T04:30:16Z","title":"RTPrune: Reading-Twice Inspired Token Pruning for Efficient DeepSeek-OCR Inference"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.00392","kind":"arxiv","version":3},"verdict":{"created_at":"2026-05-12T03:07:12.746382Z","id":"9ad105ae-5aa3-48ec-8548-c6580e248ac8","model_set":{"reader":"grok-4.3"},"one_line_summary":"RTPrune delivers 99.47% accuracy and 1.23x faster prefill on OmniDocBench for DeepSeek-OCR-Large by retaining only 84.25% of tokens through a reading-twice inspired two-stage pruning process.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"RTPrune applies a two-stage pruning process to DeepSeek-OCR that first keeps high-norm visual tokens and then merges the rest with optimal transport to cut inference time while preserving OCR accuracy.","strongest_claim":"Extensive experiments demonstrate state-of-the-art performance, as evidenced by 99.47% accuracy and 1.23× faster prefill on OmniDocBench, achieved with 84.25% token retention when applied to DeepSeek-OCR-Large.","weakest_assumption":"The observed two-stage reading trajectory in DeepSeek-OCR is stable and general enough that pruning high-norm tokens first followed by optimal-transport merging of the rest will preserve textual fidelity across OCR tasks."}},"verdict_id":"9ad105ae-5aa3-48ec-8548-c6580e248ac8"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:e0b420d86661fd78a65dc2eca3a647eb49e4c9b90a2a9f9ecf5b3bb98a1ba6b5","target":"record","created_at":"2026-05-22T01:04:03Z","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":"098170005d634e2acc0930416f394483e1c5e24fbe27629fdd2c002129c26e5e","cross_cats_sorted":["cs.LG"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2026-05-01T04:30:16Z","title_canon_sha256":"a92fc25361f723038ba98b4e0329bea793aea9d31f4909d71ddfebcc635c0a0c"},"schema_version":"1.0","source":{"id":"2605.00392","kind":"arxiv","version":3}},"canonical_sha256":"37052a58e36f9396a77613101f5d72d22f0c36fb3f6651797da4d843074d4778","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"37052a58e36f9396a77613101f5d72d22f0c36fb3f6651797da4d843074d4778","first_computed_at":"2026-05-22T01:04:03.814269Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-22T01:04:03.814269Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"N3UGgl89c0Thbew9b2B609ODApF9/rGvUd+7jt7nBiRegObxh81eH8eEIQMkiQWQ9o5IQQ/qi/BlUwSKh67qDQ==","signature_status":"signed_v1","signed_at":"2026-05-22T01:04:03.815135Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.00392","source_kind":"arxiv","source_version":3}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:e0b420d86661fd78a65dc2eca3a647eb49e4c9b90a2a9f9ecf5b3bb98a1ba6b5","sha256:3b11ff6995537b58074d3f5ba056c113de3fc654c68ca3d7d6ad12d4da7ade2b"],"state_sha256":"a25bf09d0fb7ae8a070739580911177b8f4c3981bb387de1fe5c7df09fe67cc5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"vQQj8KU6l3l3jeVL6l2gcehBwm+wXcN5PhP+uGGuaPaoDI+y4J99dMkMBQ85C0AU8qXQyHJ8jbzQONOyItwDAA==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-25T10:22:23.890663Z","bundle_sha256":"67d14ee11e3e667f57a09cf79e4da6704f4e73313dd3453587630ccf735c7031"}}