{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:ANW3I3YIY5OUNKX5IEPD5LQYE2","short_pith_number":"pith:ANW3I3YI","schema_version":"1.0","canonical_sha256":"036db46f08c75d46aafd411e3eae1826a88d1e1baa959eea4b556ba45930762b","source":{"kind":"arxiv","id":"2605.26121","version":1},"attestation_state":"computed","paper":{"title":"GEM: Geometric Entropy Mixing for Optimal LLM Data Curation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ruining Chen, Yue Min, Yujun Li, Ziyun Qiao","submitted_at":"2026-04-27T06:42:28Z","abstract_excerpt":"LLM pre-training efficacy increasingly depends on data composition rather than sheer volume. Yet, optimal mixing is hindered by categorization flaws: human taxonomies suffer from ontological misalignment, and Euclidean clustering fails to address embedding anisotropy. We introduce GEM (Geometric Entropy Mixing), a framework reformulating data curation as a variational problem on the hypersphere augmented with a mixing-balance regularizer. By decoupling the generative prior and optimizing the objective via a provable MM (Minorize-Maximize) algorithm, GEM effectively counteracts the cluster coll"},"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":"2605.26121","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2026-04-27T06:42:28Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"71bb72b00f103f0c34ec377b21536da1f768f81b71ddb914fd0d08a5862938c3","abstract_canon_sha256":"ada3234314508a6e4273a6e4001b8c2a6012a98a41e3880f841ce4483ddee932"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-27T00:04:24.847689Z","signature_b64":"QFy4cS0+/nsDPNZFR3txif91RNN7x1wnWL4rDOQBCknmV76Y0R6FXcZQLOr2nqVPOi701JtjEXRNlsyaUY20Dw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"036db46f08c75d46aafd411e3eae1826a88d1e1baa959eea4b556ba45930762b","last_reissued_at":"2026-05-27T00:04:24.846809Z","signature_status":"signed_v1","first_computed_at":"2026-05-27T00:04:24.846809Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GEM: Geometric Entropy Mixing for Optimal LLM Data Curation","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.LG","authors_text":"Ruining Chen, Yue Min, Yujun Li, Ziyun Qiao","submitted_at":"2026-04-27T06:42:28Z","abstract_excerpt":"LLM pre-training efficacy increasingly depends on data composition rather than sheer volume. Yet, optimal mixing is hindered by categorization flaws: human taxonomies suffer from ontological misalignment, and Euclidean clustering fails to address embedding anisotropy. We introduce GEM (Geometric Entropy Mixing), a framework reformulating data curation as a variational problem on the hypersphere augmented with a mixing-balance regularizer. By decoupling the generative prior and optimizing the objective via a provable MM (Minorize-Maximize) algorithm, GEM effectively counteracts the cluster coll"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2605.26121","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/2605.26121/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":"2605.26121","created_at":"2026-05-27T00:04:24.846915+00:00"},{"alias_kind":"arxiv_version","alias_value":"2605.26121v1","created_at":"2026-05-27T00:04:24.846915+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.26121","created_at":"2026-05-27T00:04:24.846915+00:00"},{"alias_kind":"pith_short_12","alias_value":"ANW3I3YIY5OU","created_at":"2026-05-27T00:04:24.846915+00:00"},{"alias_kind":"pith_short_16","alias_value":"ANW3I3YIY5OUNKX5","created_at":"2026-05-27T00:04:24.846915+00:00"},{"alias_kind":"pith_short_8","alias_value":"ANW3I3YI","created_at":"2026-05-27T00:04:24.846915+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/ANW3I3YIY5OUNKX5IEPD5LQYE2","json":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2.json","graph_json":"https://pith.science/api/pith-number/ANW3I3YIY5OUNKX5IEPD5LQYE2/graph.json","events_json":"https://pith.science/api/pith-number/ANW3I3YIY5OUNKX5IEPD5LQYE2/events.json","paper":"https://pith.science/paper/ANW3I3YI"},"agent_actions":{"view_html":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2","download_json":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2.json","view_paper":"https://pith.science/paper/ANW3I3YI","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2605.26121&json=true","fetch_graph":"https://pith.science/api/pith-number/ANW3I3YIY5OUNKX5IEPD5LQYE2/graph.json","fetch_events":"https://pith.science/api/pith-number/ANW3I3YIY5OUNKX5IEPD5LQYE2/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2/action/storage_attestation","attest_author":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2/action/author_attestation","sign_citation":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2/action/citation_signature","submit_replication":"https://pith.science/pith/ANW3I3YIY5OUNKX5IEPD5LQYE2/action/replication_record"}},"created_at":"2026-05-27T00:04:24.846915+00:00","updated_at":"2026-05-27T00:04:24.846915+00:00"}