{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:SKQLJQSC6EPULWIATGGF5RZOVQ","short_pith_number":"pith:SKQLJQSC","schema_version":"1.0","canonical_sha256":"92a0b4c242f11f45d900998c5ec72eac30b13ed6c9f32bc3de48f4e79ba65dc6","source":{"kind":"arxiv","id":"2509.02055","version":2},"attestation_state":"computed","paper":{"title":"Align-Then-stEer: Adapting the Vision-Language Action Models through Unified Latent Guidance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.RO","authors_text":"Chenjia Bai, Chenwei Wang, Chi Zhang, Ouyang Lu, Xiu Li, Xuelong Li, Yang Zhang, Yuan Zhao, Yunfei Ge, Zhenglong Sun","submitted_at":"2025-09-02T07:51:59Z","abstract_excerpt":"Vision-Language-Action (VLA) models pre-trained on large, diverse datasets show remarkable potential for general-purpose robotic manipulation. However, a primary bottleneck remains in adapting these models to downstream tasks, especially when the robot's embodiment or the task itself differs from the pre-training data. This discrepancy leads to a significant mismatch in action distributions, demanding extensive data and compute for effective fine-tuning. To address this challenge, we introduce \\textbf{Align-Then-stEer (\\texttt{ATE})}, a novel, data-efficient, and plug-and-play adaptation frame"},"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":"2509.02055","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-09-02T07:51:59Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"dbfde6e0d6561ae1a1fd5c3dfefa7fbbf516a91e0a7faef9c63552d7901b0e63","abstract_canon_sha256":"cca82dfa7c0b49f9e4dcd65c609850051f90e8b9e3721eb95a452d2edced6cb9"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T12:05:21.267421Z","signature_b64":"glZovuGPfsxSa84yvyNuXaDuWR9Jr2+4q1NKObL/nLNxJrbp4HCYISshk+u8ocNw1GXwZoN7U4ejLzTl3SEmBQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"92a0b4c242f11f45d900998c5ec72eac30b13ed6c9f32bc3de48f4e79ba65dc6","last_reissued_at":"2026-07-05T12:05:21.266928Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T12:05:21.266928Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Align-Then-stEer: Adapting the Vision-Language Action Models through Unified Latent Guidance","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.RO","authors_text":"Chenjia Bai, Chenwei Wang, Chi Zhang, Ouyang Lu, Xiu Li, Xuelong Li, Yang Zhang, Yuan Zhao, Yunfei Ge, Zhenglong Sun","submitted_at":"2025-09-02T07:51:59Z","abstract_excerpt":"Vision-Language-Action (VLA) models pre-trained on large, diverse datasets show remarkable potential for general-purpose robotic manipulation. However, a primary bottleneck remains in adapting these models to downstream tasks, especially when the robot's embodiment or the task itself differs from the pre-training data. This discrepancy leads to a significant mismatch in action distributions, demanding extensive data and compute for effective fine-tuning. To address this challenge, we introduce \\textbf{Align-Then-stEer (\\texttt{ATE})}, a novel, data-efficient, and plug-and-play adaptation frame"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2509.02055","kind":"arxiv","version":2},"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/2509.02055/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":"2509.02055","created_at":"2026-07-05T12:05:21.266982+00:00"},{"alias_kind":"arxiv_version","alias_value":"2509.02055v2","created_at":"2026-07-05T12:05:21.266982+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2509.02055","created_at":"2026-07-05T12:05:21.266982+00:00"},{"alias_kind":"pith_short_12","alias_value":"SKQLJQSC6EPU","created_at":"2026-07-05T12:05:21.266982+00:00"},{"alias_kind":"pith_short_16","alias_value":"SKQLJQSC6EPULWIA","created_at":"2026-07-05T12:05:21.266982+00:00"},{"alias_kind":"pith_short_8","alias_value":"SKQLJQSC","created_at":"2026-07-05T12:05:21.266982+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":5,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2606.12366","citing_title":"APT: Action Expert Pretraining Improves Instruction Generalization of Vision-Language-Action Policies","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2605.29562","citing_title":"VLA-Pro: Cross-Task Procedural Memory Transfer for Vision-Language-Action Models","ref_index":50,"is_internal_anchor":false},{"citing_arxiv_id":"2511.02776","citing_title":"XR-1: Towards Versatile Vision-Language-Action Models via Learning Unified Vision-Motion Representations","ref_index":103,"is_internal_anchor":false},{"citing_arxiv_id":"2511.14759","citing_title":"$\\pi^{*}_{0.6}$: a VLA That Learns From Experience","ref_index":41,"is_internal_anchor":false},{"citing_arxiv_id":"2605.04678","citing_title":"From Pixels to Tokens: A Systematic Study of Latent Action Supervision for Vision-Language-Action Models","ref_index":9,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ","json":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ.json","graph_json":"https://pith.science/api/pith-number/SKQLJQSC6EPULWIATGGF5RZOVQ/graph.json","events_json":"https://pith.science/api/pith-number/SKQLJQSC6EPULWIATGGF5RZOVQ/events.json","paper":"https://pith.science/paper/SKQLJQSC"},"agent_actions":{"view_html":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ","download_json":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ.json","view_paper":"https://pith.science/paper/SKQLJQSC","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2509.02055&json=true","fetch_graph":"https://pith.science/api/pith-number/SKQLJQSC6EPULWIATGGF5RZOVQ/graph.json","fetch_events":"https://pith.science/api/pith-number/SKQLJQSC6EPULWIATGGF5RZOVQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ/action/storage_attestation","attest_author":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ/action/author_attestation","sign_citation":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ/action/citation_signature","submit_replication":"https://pith.science/pith/SKQLJQSC6EPULWIATGGF5RZOVQ/action/replication_record"}},"created_at":"2026-07-05T12:05:21.266982+00:00","updated_at":"2026-07-05T12:05:21.266982+00:00"}