{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:5NRPO766ZMNQ52IBMLQI6XKAQF","short_pith_number":"pith:5NRPO766","schema_version":"1.0","canonical_sha256":"eb62f77fdecb1b0ee90162e08f5d40817bbf12d29b02884c2331807bc8f95c0c","source":{"kind":"arxiv","id":"2605.25842","version":1},"attestation_state":"computed","paper":{"title":"MuCRASP: Multimodal Chain-of-thought Reasoning aware Structured Pruning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Aritra Dutta, Somak Aditya","submitted_at":"2026-05-25T13:36:46Z","abstract_excerpt":"Vision-language models (VLMs) increasingly rely on chain-of-thought (CoT) reasoning to solve complex multimodal tasks, but their large parameter sizes make deployment expensive. Structured pruning offers a natural solution; however, existing methods fail to preserve CoT reasoning accuracy in VLMs. We identify two key reasons: (1) CoT consistency depends on sparse transition points (pivot tokens) in the generation trajectory, while existing pruning methods are CoT-agnostic; and (2) pruning methods designed for unimodal LLMs do not account for activation-distribution differences across visual an"},"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":"2605.25842","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.AI","submitted_at":"2026-05-25T13:36:46Z","cross_cats_sorted":["cs.CL"],"title_canon_sha256":"51e2f68fcd0263b5058dfa8d165890cd5553d5e6a122ead91a5cf701b48601f9","abstract_canon_sha256":"f3dacdd816f8329735065b37c4239702a5ff86721028967d02b8021f475331aa"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-26T02:05:14.844791Z","signature_b64":"/iuaogmhDEK7hBTVr80nfW9YYF92BpqXmMj4dZkLCSIJVmakoT9It1nQleFj3gL5xsJ/s0K47NzUhMCSY79yCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"eb62f77fdecb1b0ee90162e08f5d40817bbf12d29b02884c2331807bc8f95c0c","last_reissued_at":"2026-05-26T02:05:14.844254Z","signature_status":"signed_v1","first_computed_at":"2026-05-26T02:05:14.844254Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"MuCRASP: Multimodal Chain-of-thought Reasoning aware Structured Pruning","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CL"],"primary_cat":"cs.AI","authors_text":"Aritra Dutta, Somak Aditya","submitted_at":"2026-05-25T13:36:46Z","abstract_excerpt":"Vision-language models (VLMs) increasingly rely on chain-of-thought (CoT) reasoning to solve complex multimodal tasks, but their large parameter sizes make deployment expensive. Structured pruning offers a natural solution; however, existing methods fail to preserve CoT reasoning accuracy in VLMs. We identify two key reasons: (1) CoT consistency depends on sparse transition points (pivot tokens) in the generation trajectory, while existing pruning methods are CoT-agnostic; and (2) pruning methods designed for unimodal LLMs do not account for activation-distribution differences across visual an"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.25842","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/2605.25842/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":"2605.25842","created_at":"2026-05-26T02:05:14.844323+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.25842v1","created_at":"2026-05-26T02:05:14.844323+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.25842","created_at":"2026-05-26T02:05:14.844323+00:00"},{"alias_kind":"pith_short_12","alias_value":"5NRPO766ZMNQ","created_at":"2026-05-26T02:05:14.844323+00:00"},{"alias_kind":"pith_short_16","alias_value":"5NRPO766ZMNQ52IB","created_at":"2026-05-26T02:05:14.844323+00:00"},{"alias_kind":"pith_short_8","alias_value":"5NRPO766","created_at":"2026-05-26T02:05:14.844323+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":0,"internal_anchor_count":0,"sample":[]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF","json":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF.json","graph_json":"https://pith.science/api/pith-number/5NRPO766ZMNQ52IBMLQI6XKAQF/graph.json","events_json":"https://pith.science/api/pith-number/5NRPO766ZMNQ52IBMLQI6XKAQF/events.json","paper":"https://pith.science/paper/5NRPO766"},"agent_actions":{"view_html":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF","download_json":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF.json","view_paper":"https://pith.science/paper/5NRPO766","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.25842&json=true","fetch_graph":"https://pith.science/api/pith-number/5NRPO766ZMNQ52IBMLQI6XKAQF/graph.json","fetch_events":"https://pith.science/api/pith-number/5NRPO766ZMNQ52IBMLQI6XKAQF/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF/action/timestamp_anchor","attest_storage":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF/action/storage_attestation","attest_author":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF/action/author_attestation","sign_citation":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF/action/citation_signature","submit_replication":"https://pith.science/pith/5NRPO766ZMNQ52IBMLQI6XKAQF/action/replication_record"}},"created_at":"2026-05-26T02:05:14.844323+00:00","updated_at":"2026-05-26T02:05:14.844323+00:00"}