{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MSLSR27S7Q75SXJX35OZHK5WYI","short_pith_number":"pith:MSLSR27S","schema_version":"1.0","canonical_sha256":"649728ebf2fc3fd95d37df5d93abb6c200f6d5abe002cc883edd9c36e644c261","source":{"kind":"arxiv","id":"2509.09121","version":1},"attestation_state":"computed","paper":{"title":"Compass-v3: Scaling Domain-Specific LLMs for Multilingual E-Commerce in Southeast Asia","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Sophia Maria","submitted_at":"2025-09-11T03:23:48Z","abstract_excerpt":"Large language models (LLMs) excel in general-domain applications, yet their performance often degrades in specialized tasks requiring domain-specific knowledge. E-commerce is particularly challenging, as its data are noisy, heterogeneous, multilingual, and highly dynamic. We present Compass-v3, a vertical-domain Mixture-of-Experts (MoE) model with 245B total parameters and 71B active per token, designed for Southeast Asian e-commerce. Compass-v3 adopts fewer but larger experts, combined with hardware-efficient optimizations-such as intra-node expert parallelism and a customized memcpy operato"},"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":"2509.09121","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2025-09-11T03:23:48Z","cross_cats_sorted":[],"title_canon_sha256":"af24a93ab9af838df1db90200ccc5e8c74b57be19a891f0f556d54cdde851ec6","abstract_canon_sha256":"3a607d17dcc551573eeb15e088f8d8cee2b3fc0b35384640ed9ad4fc38b02e4e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:09:29.899843Z","signature_b64":"hit1UPO0Cbyn6fkpHep4KhcbmMls8kb6MBc39eUiQaHMJcYJhwv0i1FvI797fRYw/sR/yZbRRNSmsguryPYYCQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"649728ebf2fc3fd95d37df5d93abb6c200f6d5abe002cc883edd9c36e644c261","last_reissued_at":"2026-07-05T12:09:29.899289Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:09:29.899289Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Compass-v3: Scaling Domain-Specific LLMs for Multilingual E-Commerce in Southeast Asia","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CL","authors_text":"Sophia Maria","submitted_at":"2025-09-11T03:23:48Z","abstract_excerpt":"Large language models (LLMs) excel in general-domain applications, yet their performance often degrades in specialized tasks requiring domain-specific knowledge. E-commerce is particularly challenging, as its data are noisy, heterogeneous, multilingual, and highly dynamic. We present Compass-v3, a vertical-domain Mixture-of-Experts (MoE) model with 245B total parameters and 71B active per token, designed for Southeast Asian e-commerce. Compass-v3 adopts fewer but larger experts, combined with hardware-efficient optimizations-such as intra-node expert parallelism and a customized memcpy operato"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.09121","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/2509.09121/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":"2509.09121","created_at":"2026-07-05T12:09:29.899356+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.09121v1","created_at":"2026-07-05T12:09:29.899356+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.09121","created_at":"2026-07-05T12:09:29.899356+00:00"},{"alias_kind":"pith_short_12","alias_value":"MSLSR27S7Q75","created_at":"2026-07-05T12:09:29.899356+00:00"},{"alias_kind":"pith_short_16","alias_value":"MSLSR27S7Q75SXJX","created_at":"2026-07-05T12:09:29.899356+00:00"},{"alias_kind":"pith_short_8","alias_value":"MSLSR27S","created_at":"2026-07-05T12:09:29.899356+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":3,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.28070","citing_title":"JD Oxygen AI Item Center (Oxygen AIIC) V1: An Industrial-Scale LLM/VLM-Centric Solution for Item Understanding, Management, and Applications","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.28070","citing_title":"JD Oxygen AI Item Center (Oxygen AIIC) V1: An Industrial-Scale LLM/VLM-Centric Solution for Item Understanding, Management, and Applications","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01278","citing_title":"Valley3: Scaling Omni Foundation Models for E-commerce","ref_index":47,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI","json":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI.json","graph_json":"https://pith.science/api/pith-number/MSLSR27S7Q75SXJX35OZHK5WYI/graph.json","events_json":"https://pith.science/api/pith-number/MSLSR27S7Q75SXJX35OZHK5WYI/events.json","paper":"https://pith.science/paper/MSLSR27S"},"agent_actions":{"view_html":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI","download_json":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI.json","view_paper":"https://pith.science/paper/MSLSR27S","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.09121&json=true","fetch_graph":"https://pith.science/api/pith-number/MSLSR27S7Q75SXJX35OZHK5WYI/graph.json","fetch_events":"https://pith.science/api/pith-number/MSLSR27S7Q75SXJX35OZHK5WYI/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI/action/storage_attestation","attest_author":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI/action/author_attestation","sign_citation":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI/action/citation_signature","submit_replication":"https://pith.science/pith/MSLSR27S7Q75SXJX35OZHK5WYI/action/replication_record"}},"created_at":"2026-07-05T12:09:29.899356+00:00","updated_at":"2026-07-05T12:09:29.899356+00:00"}