{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:MRK7GAQ7MISIY5PYX4LELZZHBS","short_pith_number":"pith:MRK7GAQ7","schema_version":"1.0","canonical_sha256":"6455f3021f62248c75f8bf1645e7270c95ba7afd3bde999c2a5c763ad4d278f5","source":{"kind":"arxiv","id":"2502.07903","version":1},"attestation_state":"computed","paper":{"title":"HexGen-2: Disaggregated Generative Inference of LLMs in Heterogeneous Environment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Binhang Yuan, Ran Yan, Youhe Jiang","submitted_at":"2025-02-11T19:17:35Z","abstract_excerpt":"Disaggregating the prefill and decoding phases represents an effective new paradigm for generative inference of large language models (LLM), which eliminates prefill-decoding interference and optimizes resource allocation. However, it is still an open problem about how to deploy the disaggregated inference paradigm across a group of heterogeneous GPUs, which can be an economical alternative to deployment over homogeneous high-performance GPUs. Towards this end, we introduce HexGen-2, a distributed system for efficient and economical LLM serving on heterogeneous GPUs following the disaggregated"},"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":"2502.07903","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.DC","submitted_at":"2025-02-11T19:17:35Z","cross_cats_sorted":[],"title_canon_sha256":"043ca7c7212bea4016867240352a203636f35ee6df914827bdddfdd936c5b8aa","abstract_canon_sha256":"17df8fee36803d4714f19ed0989dc753e4df55ffd657ad5380b61adf7aa02df8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:13:11.893300Z","signature_b64":"Vl7B3G+I2gOVAdcoLgtCvzJhGsuVwkY1ZcQaixOJs7dsd3zcSaK/PoC6Z7H47/K62DmoOtT7Eh1BRswrcLCABA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6455f3021f62248c75f8bf1645e7270c95ba7afd3bde999c2a5c763ad4d278f5","last_reissued_at":"2026-07-05T10:13:11.892818Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:13:11.892818Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"HexGen-2: Disaggregated Generative Inference of LLMs in Heterogeneous Environment","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.DC","authors_text":"Binhang Yuan, Ran Yan, Youhe Jiang","submitted_at":"2025-02-11T19:17:35Z","abstract_excerpt":"Disaggregating the prefill and decoding phases represents an effective new paradigm for generative inference of large language models (LLM), which eliminates prefill-decoding interference and optimizes resource allocation. However, it is still an open problem about how to deploy the disaggregated inference paradigm across a group of heterogeneous GPUs, which can be an economical alternative to deployment over homogeneous high-performance GPUs. Towards this end, we introduce HexGen-2, a distributed system for efficient and economical LLM serving on heterogeneous GPUs following the disaggregated"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2502.07903","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/2502.07903/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":"2502.07903","created_at":"2026-07-05T10:13:11.892874+00:00"},{"alias_kind":"arxiv_version","alias_value":"2502.07903v1","created_at":"2026-07-05T10:13:11.892874+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2502.07903","created_at":"2026-07-05T10:13:11.892874+00:00"},{"alias_kind":"pith_short_12","alias_value":"MRK7GAQ7MISI","created_at":"2026-07-05T10:13:11.892874+00:00"},{"alias_kind":"pith_short_16","alias_value":"MRK7GAQ7MISIY5PY","created_at":"2026-07-05T10:13:11.892874+00:00"},{"alias_kind":"pith_short_8","alias_value":"MRK7GAQ7","created_at":"2026-07-05T10:13:11.892874+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":6,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.19271","citing_title":"TurboServe: Serving Streaming Video Generation Efficiently and Economically","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2409.01143","citing_title":"HexiScale: Facilitating Large Language Model Training over Heterogeneous Hardware","ref_index":21,"is_internal_anchor":false},{"citing_arxiv_id":"2605.16637","citing_title":"HexAGenT: Efficient Agentic LLM Serving via Workflow- and Heterogeneity-Aware Scheduling","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.10180","citing_title":"Tessera: Unlocking Heterogeneous GPUs through Kernel-Granularity Disaggregation","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07144","citing_title":"Autopoiesis: A Self-Evolving System Paradigm for LLM Serving Under Runtime Dynamics","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07569","citing_title":"HexiSeq: Accommodating Long Context Training of LLMs over Heterogeneous Hardware","ref_index":24,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS","json":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS.json","graph_json":"https://pith.science/api/pith-number/MRK7GAQ7MISIY5PYX4LELZZHBS/graph.json","events_json":"https://pith.science/api/pith-number/MRK7GAQ7MISIY5PYX4LELZZHBS/events.json","paper":"https://pith.science/paper/MRK7GAQ7"},"agent_actions":{"view_html":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS","download_json":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS.json","view_paper":"https://pith.science/paper/MRK7GAQ7","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2502.07903&json=true","fetch_graph":"https://pith.science/api/pith-number/MRK7GAQ7MISIY5PYX4LELZZHBS/graph.json","fetch_events":"https://pith.science/api/pith-number/MRK7GAQ7MISIY5PYX4LELZZHBS/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS/action/timestamp_anchor","attest_storage":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS/action/storage_attestation","attest_author":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS/action/author_attestation","sign_citation":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS/action/citation_signature","submit_replication":"https://pith.science/pith/MRK7GAQ7MISIY5PYX4LELZZHBS/action/replication_record"}},"created_at":"2026-07-05T10:13:11.892874+00:00","updated_at":"2026-07-05T10:13:11.892874+00:00"}