{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ZNJLEU3FMITO6OMG33M45FJRJG","short_pith_number":"pith:ZNJLEU3F","schema_version":"1.0","canonical_sha256":"cb52b253656226ef3986ded9ce953149b4f42b78bbc25fff0e3d7df4c95dce7b","source":{"kind":"arxiv","id":"2608.07814","version":1},"attestation_state":"computed","paper":{"title":"Shape Mutating Expert Compression:LorExperts and BTExperts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ashish Sirasao, Bowen Bao, Inesh Chakrabarti, Sourjya Roy, Spandan Tiwari, Thiago Crepaldi","submitted_at":"2026-08-07T23:23:40Z","abstract_excerpt":"Mixture-of-Experts (MoE) language models deliver high capacity at low per-token compute, but deploying them cheaply requires compressing their many expert weight matrices. Expert pruning (e.g., REAP) and merging reduce cost but sacrifice accuracy and require retraining the router; low-rank delta decomposition of experts (e.g., D^2-MoE) preserves all experts and the router, but degrades sharply as the expert count grows because a single shared component cannot approximate many near-orthogonal experts.\n  Because MoE expert weights are near-orthogonal, a single shared component (as in prior delta"},"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":"2608.07814","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-08-07T23:23:40Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"97032995c1e470487a7715cce4ff40e09a2c0b4090bb21e923e100dfcef69ec3","abstract_canon_sha256":"4436bb85b22856429e520337aa0605e7a534a1b70cb76a9313f46b99f27bef16"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-11T01:18:21.173097Z","signature_b64":"rwsvbVueFss0zFpZVzp3TGzt8ttxfXQJV/RyOoOH8xF2+hObqyszTLOBt1KNCY0qM9HdPEIfgt/sf2+R1pGQBA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"cb52b253656226ef3986ded9ce953149b4f42b78bbc25fff0e3d7df4c95dce7b","last_reissued_at":"2026-08-11T01:18:21.167317Z","signature_status":"signed_v1","first_computed_at":"2026-08-11T01:18:21.167317Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Shape Mutating Expert Compression:LorExperts and BTExperts","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ashish Sirasao, Bowen Bao, Inesh Chakrabarti, Sourjya Roy, Spandan Tiwari, Thiago Crepaldi","submitted_at":"2026-08-07T23:23:40Z","abstract_excerpt":"Mixture-of-Experts (MoE) language models deliver high capacity at low per-token compute, but deploying them cheaply requires compressing their many expert weight matrices. Expert pruning (e.g., REAP) and merging reduce cost but sacrifice accuracy and require retraining the router; low-rank delta decomposition of experts (e.g., D^2-MoE) preserves all experts and the router, but degrades sharply as the expert count grows because a single shared component cannot approximate many near-orthogonal experts.\n  Because MoE expert weights are near-orthogonal, a single shared component (as in prior delta"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.07814","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/2608.07814/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":"2608.07814","created_at":"2026-08-11T01:18:21.171215+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.07814v1","created_at":"2026-08-11T01:18:21.171215+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.07814","created_at":"2026-08-11T01:18:21.171215+00:00"},{"alias_kind":"pith_short_12","alias_value":"ZNJLEU3FMITO","created_at":"2026-08-11T01:18:21.171215+00:00"},{"alias_kind":"pith_short_16","alias_value":"ZNJLEU3FMITO6OMG","created_at":"2026-08-11T01:18:21.171215+00:00"},{"alias_kind":"pith_short_8","alias_value":"ZNJLEU3F","created_at":"2026-08-11T01:18:21.171215+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/ZNJLEU3FMITO6OMG33M45FJRJG","json":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG.json","graph_json":"https://pith.science/api/pith-number/ZNJLEU3FMITO6OMG33M45FJRJG/graph.json","events_json":"https://pith.science/api/pith-number/ZNJLEU3FMITO6OMG33M45FJRJG/events.json","paper":"https://pith.science/paper/ZNJLEU3F"},"agent_actions":{"view_html":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG","download_json":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG.json","view_paper":"https://pith.science/paper/ZNJLEU3F","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.07814&json=true","fetch_graph":"https://pith.science/api/pith-number/ZNJLEU3FMITO6OMG33M45FJRJG/graph.json","fetch_events":"https://pith.science/api/pith-number/ZNJLEU3FMITO6OMG33M45FJRJG/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG/action/storage_attestation","attest_author":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG/action/author_attestation","sign_citation":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG/action/citation_signature","submit_replication":"https://pith.science/pith/ZNJLEU3FMITO6OMG33M45FJRJG/action/replication_record"}},"created_at":"2026-08-11T01:18:21.171215+00:00","updated_at":"2026-08-11T01:18:21.171215+00:00"}