{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:OOWRA376HPHIRDXBT3CXHTFH5W","short_pith_number":"pith:OOWRA376","schema_version":"1.0","canonical_sha256":"73ad106ffe3bce888ee19ec573cca7ed9e35ec52e3799d1bca20a5aa5af19c43","source":{"kind":"arxiv","id":"2607.22614","version":1},"attestation_state":"computed","paper":{"title":"DynaResize: Runtime GPU Reallocation for Disaggregated LLM Post-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Haiquan Chen, Hanlin Du, Jiarui Fang, Sa Wang, Yungang Bao, Zhiyuan Yan","submitted_at":"2026-06-15T09:41:09Z","abstract_excerpt":"RL-based LLM post-training increasingly disaggregates Rollout and Training across separate GPU resources, but static GPU partitioning suffers from severe pipeline bubbles under long-tail rollout latency. We present DynaResize, a runtime GPU reallocation system that dynamically switches GPUs between Rollout and Training to balance stage execution times without changing RL semantics. DynaResize decomposes resizing into fine-grained operations and removes non-startup-critical work from the critical path through communicator reuse, bounded state staging, and hysteresis-based resizing. Experimental"},"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":"2607.22614","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.AI","submitted_at":"2026-06-15T09:41:09Z","cross_cats_sorted":[],"title_canon_sha256":"46a710afedebea4cc55a0499efcc4a5e14c3ae97593557dd474df4888fe88933","abstract_canon_sha256":"c9097661772a046ce142d91e9192a4193ebc157ffae91225a53cadabc9f09a1e"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-28T00:21:45.125985Z","signature_b64":"HlLg+9PXiTAxYnSD/W4a+L8H0EwVuXmz7jiPGPqGL4d2xjd0vepPm/BsqrHLoie0NpSCBD8pi54eovLoy8V4CQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"73ad106ffe3bce888ee19ec573cca7ed9e35ec52e3799d1bca20a5aa5af19c43","last_reissued_at":"2026-07-28T00:21:45.124963Z","signature_status":"signed_v1","first_computed_at":"2026-07-28T00:21:45.124963Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"DynaResize: Runtime GPU Reallocation for Disaggregated LLM Post-Training","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.AI","authors_text":"Haiquan Chen, Hanlin Du, Jiarui Fang, Sa Wang, Yungang Bao, Zhiyuan Yan","submitted_at":"2026-06-15T09:41:09Z","abstract_excerpt":"RL-based LLM post-training increasingly disaggregates Rollout and Training across separate GPU resources, but static GPU partitioning suffers from severe pipeline bubbles under long-tail rollout latency. We present DynaResize, a runtime GPU reallocation system that dynamically switches GPUs between Rollout and Training to balance stage execution times without changing RL semantics. DynaResize decomposes resizing into fine-grained operations and removes non-startup-critical work from the critical path through communicator reuse, bounded state staging, and hysteresis-based resizing. Experimental"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2607.22614","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/2607.22614/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":"2607.22614","created_at":"2026-07-28T00:21:45.125555+00:00"},{"alias_kind":"arxiv_version","alias_value":"2607.22614v1","created_at":"2026-07-28T00:21:45.125555+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2607.22614","created_at":"2026-07-28T00:21:45.125555+00:00"},{"alias_kind":"pith_short_12","alias_value":"OOWRA376HPHI","created_at":"2026-07-28T00:21:45.125555+00:00"},{"alias_kind":"pith_short_16","alias_value":"OOWRA376HPHIRDXB","created_at":"2026-07-28T00:21:45.125555+00:00"},{"alias_kind":"pith_short_8","alias_value":"OOWRA376","created_at":"2026-07-28T00:21:45.125555+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/OOWRA376HPHIRDXBT3CXHTFH5W","json":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W.json","graph_json":"https://pith.science/api/pith-number/OOWRA376HPHIRDXBT3CXHTFH5W/graph.json","events_json":"https://pith.science/api/pith-number/OOWRA376HPHIRDXBT3CXHTFH5W/events.json","paper":"https://pith.science/paper/OOWRA376"},"agent_actions":{"view_html":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W","download_json":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W.json","view_paper":"https://pith.science/paper/OOWRA376","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2607.22614&json=true","fetch_graph":"https://pith.science/api/pith-number/OOWRA376HPHIRDXBT3CXHTFH5W/graph.json","fetch_events":"https://pith.science/api/pith-number/OOWRA376HPHIRDXBT3CXHTFH5W/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W/action/timestamp_anchor","attest_storage":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W/action/storage_attestation","attest_author":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W/action/author_attestation","sign_citation":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W/action/citation_signature","submit_replication":"https://pith.science/pith/OOWRA376HPHIRDXBT3CXHTFH5W/action/replication_record"}},"created_at":"2026-07-28T00:21:45.125555+00:00","updated_at":"2026-07-28T00:21:45.125555+00:00"}