{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:Z6HLR2KOITW2GTPLADW3PQM6WY","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":"d89c916e7a11a6617f46c67f171b91b726f96ccf276a611b3aeeebd5f145ee1e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-14T03:22:30Z","title_canon_sha256":"0b3c0d18e1f31bb8b3496673bb86fa0d1b96f8140acf5c2ebf52d0e4339d3520"},"schema_version":"1.0","source":{"id":"2605.14310","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.14310","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"arxiv_version","alias_value":"2605.14310v1","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.14310","created_at":"2026-05-17T23:39:10Z"},{"alias_kind":"pith_short_12","alias_value":"Z6HLR2KOITW2","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"Z6HLR2KOITW2GTPL","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"Z6HLR2KO","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:ba582acfeeeb770fac18d3c64ac1980f536e0897c92f25ce5ee1e0afb2dc1a08","target":"graph","created_at":"2026-05-17T23:39:10Z","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":"Across four open-source VLMs and five long-video and streaming-video benchmarks, our method improves over heuristic streaming compression baselines under a fixed cache budget."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That a small coreset chosen by joint KV coverage and orthogonality will preserve the information needed for arbitrary future reasoning queries over the full history."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"CoRDS selects a compact KV-cache subset via joint-space coreset coverage and log-det diversity to outperform token-wise heuristics on long-video VLM benchmarks."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Treating KV-cache compression as coreset selection improves streaming video understanding under fixed memory budgets."}],"snapshot_sha256":"d3c1de20f15168de3249dfe5176e72b4d1eda4bd635eb4d9c36fff85a1725ba3"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"b1cc6b3b748e09ee0735fa762b4d42b27c47d6e4fc29ccb0e729e336af5b5524"},"paper":{"abstract_excerpt":"Streaming video understanding with large vision-language models (VLMs) requires a compact memory that can support future reasoning over an ever-growing visual history. A common solution is to compress the key-value (KV) cache, but existing streaming methods typically rely on local token-wise heuristics, such as recency, temporal redundancy, or saliency, which do not explicitly optimize whether the retained cache is representative of the accumulated history. We propose to view KV-cache compression as a coreset selection problem: rather than scoring tokens independently for retention, we select ","authors_text":"Ailar Mahdizadeh, Leonid Sigal, Muchen Li, Puria Azadi, Xiangteng He","cross_cats":[],"headline":"Treating KV-cache compression as coreset selection improves streaming video understanding under fixed memory budgets.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-14T03:22:30Z","title":"CoRDS: Coreset-based Representative and Diverse Selection for Streaming Video Understanding"},"references":{"count":32,"internal_anchors":7,"resolved_work":32,"sample":[{"cited_arxiv_id":"2409.12191","doi":"","is_internal_anchor":true,"ref_index":2,"title":"Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution","work_id":"8abcfe4f-e0fb-44b7-9123-448fac95f90a","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Long context transfer from language to vision.Transactions on Machine Learning Research, 2025","work_id":"62705864-363c-4f27-84b3-0f90e70f58f2","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Infinipot-v: Memory-constrained kv cache compression for streaming video understanding","work_id":"fd86124e-903d-4c8e-a169-f9b97c80b4b6","year":2025},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Streaming long video understanding with large language models","work_id":"6a0c4f9b-d588-49db-8fe1-5d2652cc34f5","year":2024},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":6,"title":"Kim, Bilge Soran, Raghuraman Krishnamoorthi, Mohamed Elhoseiny, and Vikas Chandra","work_id":"e47963a0-7038-4696-99ee-74c58cd4e5de","year":2025}],"snapshot_sha256":"e47305bb4df4f97304fbc7b8522295edd67bd1fc08b7cc0c177e4d4cab054d3f"},"source":{"id":"2605.14310","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-15T02:24:37.272302Z","id":"4a664be5-ccb1-4be9-a9c7-1ea9c582dcb5","model_set":{"reader":"grok-4.3"},"one_line_summary":"CoRDS selects a compact KV-cache subset via joint-space coreset coverage and log-det diversity to outperform token-wise heuristics on long-video VLM benchmarks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Treating KV-cache compression as coreset selection improves streaming video understanding under fixed memory budgets.","strongest_claim":"Across four open-source VLMs and five long-video and streaming-video benchmarks, our method improves over heuristic streaming compression baselines under a fixed cache budget.","weakest_assumption":"That a small coreset chosen by joint KV coverage and orthogonality will preserve the information needed for arbitrary future reasoning queries over the full history."}},"verdict_id":"4a664be5-ccb1-4be9-a9c7-1ea9c582dcb5"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:7e230e622bc4ff3043e63ec0178f398b809818c1c656f9c3162a7b4b29b40832","target":"record","created_at":"2026-05-17T23:39:10Z","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":"d89c916e7a11a6617f46c67f171b91b726f96ccf276a611b3aeeebd5f145ee1e","cross_cats_sorted":[],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2026-05-14T03:22:30Z","title_canon_sha256":"0b3c0d18e1f31bb8b3496673bb86fa0d1b96f8140acf5c2ebf52d0e4339d3520"},"schema_version":"1.0","source":{"id":"2605.14310","kind":"arxiv","version":1}},"canonical_sha256":"cf8eb8e94e44eda34deb00edb7c19eb6111e73e74512c24e1287abf96d159910","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"cf8eb8e94e44eda34deb00edb7c19eb6111e73e74512c24e1287abf96d159910","first_computed_at":"2026-05-17T23:39:10.003699Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:10.003699Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"Q5LWr7kRKDnM4L0iz9/m2xo6DH6QWWlflH6CLUz9yk3bsB/aBGm7oBy7vUqbi3uon4zMkXql22Kbn4EWFoxlBg==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:10.004159Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.14310","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:7e230e622bc4ff3043e63ec0178f398b809818c1c656f9c3162a7b4b29b40832","sha256:ba582acfeeeb770fac18d3c64ac1980f536e0897c92f25ce5ee1e0afb2dc1a08"],"state_sha256":"b34e266b982bddd18acf5f3d7f5025d0bc5d056428f9f2223f01f75590f7beb6"}