{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:CHRL362KGOIVLBTRFVXTYQML55","short_pith_number":"pith:CHRL362K","schema_version":"1.0","canonical_sha256":"11e2bdfb4a33915586712d6f3c418bef4f177a14e2422f60cd1dc431fbd66d16","source":{"kind":"arxiv","id":"2503.05641","version":4},"attestation_state":"computed","paper":{"title":"Skill-Based Mixture-of-Experts: Adaptive Routing for Heterogeneous Reasoning via Inferred Skills","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Elias Stengel-Eskin, Justin Chih-Yao Chen, Mohit Bansal, Sukwon Yun, Tianlong Chen","submitted_at":"2025-03-07T18:03:13Z","abstract_excerpt":"Combining existing pre-trained LLMs is a promising approach for diverse reasoning tasks. However, task-level expert selection is often too coarse-grained, since different instances may require different expertise. To address this, we propose Skill-MoE, a symbolic, skill-based, and gradient-free Mixture-of-Experts framework for instance-level expert selection. Skill-MoE infers skills (e.g., algebra in mathematics) from each query, selects experts based on skill relevance, and lets each expert generate its own reasoning. The resulting k outputs are then synthesized by an aggregator chosen for it"},"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":"2503.05641","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-03-07T18:03:13Z","cross_cats_sorted":["cs.AI","cs.LG"],"title_canon_sha256":"1dd38a71765826da339f99890e4b16c24f1b639ba9c15fa63b9963dc63b5b2ba","abstract_canon_sha256":"9d70d64113436fc5b2ce760585f5638f3fef53d47e69a9882bf5bf8564323bbe"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-06-02T01:03:30.681081Z","signature_b64":"V9sHVAtWgHAnD9dA9mZM4T9XKM8HH8OgoGCBKNFmUCgJVt5TYMFK2BlQZELc6drYPsDpCUqGvKu65j5TDYFwCg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"11e2bdfb4a33915586712d6f3c418bef4f177a14e2422f60cd1dc431fbd66d16","last_reissued_at":"2026-06-02T01:03:30.680539Z","signature_status":"signed_v1","first_computed_at":"2026-06-02T01:03:30.680539Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Skill-Based Mixture-of-Experts: Adaptive Routing for Heterogeneous Reasoning via Inferred Skills","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Elias Stengel-Eskin, Justin Chih-Yao Chen, Mohit Bansal, Sukwon Yun, Tianlong Chen","submitted_at":"2025-03-07T18:03:13Z","abstract_excerpt":"Combining existing pre-trained LLMs is a promising approach for diverse reasoning tasks. However, task-level expert selection is often too coarse-grained, since different instances may require different expertise. To address this, we propose Skill-MoE, a symbolic, skill-based, and gradient-free Mixture-of-Experts framework for instance-level expert selection. Skill-MoE infers skills (e.g., algebra in mathematics) from each query, selects experts based on skill relevance, and lets each expert generate its own reasoning. The resulting k outputs are then synthesized by an aggregator chosen for it"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.05641","kind":"arxiv","version":4},"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/2503.05641/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":"2503.05641","created_at":"2026-06-02T01:03:30.680604+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.05641v4","created_at":"2026-06-02T01:03:30.680604+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.05641","created_at":"2026-06-02T01:03:30.680604+00:00"},{"alias_kind":"pith_short_12","alias_value":"CHRL362KGOIV","created_at":"2026-06-02T01:03:30.680604+00:00"},{"alias_kind":"pith_short_16","alias_value":"CHRL362KGOIVLBTR","created_at":"2026-06-02T01:03:30.680604+00:00"},{"alias_kind":"pith_short_8","alias_value":"CHRL362K","created_at":"2026-06-02T01:03:30.680604+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":3,"sample":[{"citing_arxiv_id":"2507.14200","citing_title":"A Scalable Multi-LLM Collaboration System with Retrieval-based Selection and Exploration-Exploitation-Driven Enhancement","ref_index":13,"is_internal_anchor":true},{"citing_arxiv_id":"2605.17486","citing_title":"DyGRO-VLA: Cross-Task Scaling of Vision-Language-Action Models via Dynamic Grouped Residual Optimization","ref_index":119,"is_internal_anchor":true},{"citing_arxiv_id":"2605.04069","citing_title":"LAWS: Learning from Actual Workloads Symbolically -- A Self-Certifying Parametrized Cache Architecture for Neural Inference, Robotics, and Edge Deployment","ref_index":21,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55","json":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55.json","graph_json":"https://pith.science/api/pith-number/CHRL362KGOIVLBTRFVXTYQML55/graph.json","events_json":"https://pith.science/api/pith-number/CHRL362KGOIVLBTRFVXTYQML55/events.json","paper":"https://pith.science/paper/CHRL362K"},"agent_actions":{"view_html":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55","download_json":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55.json","view_paper":"https://pith.science/paper/CHRL362K","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.05641&json=true","fetch_graph":"https://pith.science/api/pith-number/CHRL362KGOIVLBTRFVXTYQML55/graph.json","fetch_events":"https://pith.science/api/pith-number/CHRL362KGOIVLBTRFVXTYQML55/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55/action/storage_attestation","attest_author":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55/action/author_attestation","sign_citation":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55/action/citation_signature","submit_replication":"https://pith.science/pith/CHRL362KGOIVLBTRFVXTYQML55/action/replication_record"}},"created_at":"2026-06-02T01:03:30.680604+00:00","updated_at":"2026-06-02T01:03:30.680604+00:00"}