{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2026:WTIZHG2PV2O5YPS5QMP2YO2M7G","short_pith_number":"pith:WTIZHG2P","schema_version":"1.0","canonical_sha256":"b4d1939b4fae9ddc3e5d831fac3b4cf9ae8d39db8404eab575641b39aff2dc07","source":{"kind":"arxiv","id":"2608.07314","version":1},"attestation_state":"computed","paper":{"title":"TEMPO: Semantic-Action Decoupled RL Post-Training for Vision-Language-Action Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.RO","authors_text":"Quantao Yang, Ziheng Liu","submitted_at":"2026-08-07T15:09:51Z","abstract_excerpt":"Vision-language-action (VLA) models are commonly adapted to downstream manipulation tasks via supervised fine-tuning (SFT) or online reinforcement learning (RL) post-training. SFT is prone to distribution mismatch, and existing RL approaches typically apply a single, uniform update strategy to all model components, ignoring their distinct functional roles. We propose TEMPO, a semantic-action decoupled, two-timescale RL post-training framework for VLA models. TEMPO freezes the pretrained vision-language backbone to preserve general semantic representations, and restricts adaptation to two compo"},"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":"2608.07314","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2026-08-07T15:09:51Z","cross_cats_sorted":["cs.CV"],"title_canon_sha256":"1693529a50d01de14916fac4ba0acec9dc3c29c7c62eda5ae05a3201308b017e","abstract_canon_sha256":"625209d19b75747078c4f387341901749394330e41f227543e507bd1451cc8e8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-10T01:14:57.549419Z","signature_b64":"k57RWVEjo26fasMY/gsKS+onXLB/nA3QU4F6PV2WFH6t00Xrs0HWPBjzcdg3nuSUKyKX4+mgY+2l2nsZ9fi4Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"b4d1939b4fae9ddc3e5d831fac3b4cf9ae8d39db8404eab575641b39aff2dc07","last_reissued_at":"2026-08-10T01:14:57.547178Z","signature_status":"signed_v1","first_computed_at":"2026-08-10T01:14:57.547178Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"TEMPO: Semantic-Action Decoupled RL Post-Training for Vision-Language-Action Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.CV"],"primary_cat":"cs.RO","authors_text":"Quantao Yang, Ziheng Liu","submitted_at":"2026-08-07T15:09:51Z","abstract_excerpt":"Vision-language-action (VLA) models are commonly adapted to downstream manipulation tasks via supervised fine-tuning (SFT) or online reinforcement learning (RL) post-training. SFT is prone to distribution mismatch, and existing RL approaches typically apply a single, uniform update strategy to all model components, ignoring their distinct functional roles. We propose TEMPO, a semantic-action decoupled, two-timescale RL post-training framework for VLA models. TEMPO freezes the pretrained vision-language backbone to preserve general semantic representations, and restricts adaptation to two compo"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.07314","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/2608.07314/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":"2608.07314","created_at":"2026-08-10T01:14:57.548064+00:00"},{"alias_kind":"arxiv_version","alias_value":"2608.07314v1","created_at":"2026-08-10T01:14:57.548064+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.07314","created_at":"2026-08-10T01:14:57.548064+00:00"},{"alias_kind":"pith_short_12","alias_value":"WTIZHG2PV2O5","created_at":"2026-08-10T01:14:57.548064+00:00"},{"alias_kind":"pith_short_16","alias_value":"WTIZHG2PV2O5YPS5","created_at":"2026-08-10T01:14:57.548064+00:00"},{"alias_kind":"pith_short_8","alias_value":"WTIZHG2P","created_at":"2026-08-10T01:14:57.548064+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/WTIZHG2PV2O5YPS5QMP2YO2M7G","json":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G.json","graph_json":"https://pith.science/api/pith-number/WTIZHG2PV2O5YPS5QMP2YO2M7G/graph.json","events_json":"https://pith.science/api/pith-number/WTIZHG2PV2O5YPS5QMP2YO2M7G/events.json","paper":"https://pith.science/paper/WTIZHG2P"},"agent_actions":{"view_html":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G","download_json":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G.json","view_paper":"https://pith.science/paper/WTIZHG2P","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2608.07314&json=true","fetch_graph":"https://pith.science/api/pith-number/WTIZHG2PV2O5YPS5QMP2YO2M7G/graph.json","fetch_events":"https://pith.science/api/pith-number/WTIZHG2PV2O5YPS5QMP2YO2M7G/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G/action/timestamp_anchor","attest_storage":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G/action/storage_attestation","attest_author":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G/action/author_attestation","sign_citation":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G/action/citation_signature","submit_replication":"https://pith.science/pith/WTIZHG2PV2O5YPS5QMP2YO2M7G/action/replication_record"}},"created_at":"2026-08-10T01:14:57.548064+00:00","updated_at":"2026-08-10T01:14:57.548064+00:00"}