{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:55FVLLS257ANMDE2F5DU2B3ESM","short_pith_number":"pith:55FVLLS2","canonical_record":{"source":{"id":"2605.13382","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2026-05-13T11:37:51Z","cross_cats_sorted":[],"title_canon_sha256":"b117477d7afc8cf4c8c6746c53e675be06c5f931688f7bea381ae57838da0a66","abstract_canon_sha256":"ecb922ceebb7025b32744b266b68f0d834cf5fc491a3a208020e76698e38da41"},"schema_version":"1.0"},"canonical_sha256":"ef4b55ae5aefc0d60c9a2f474d0764930420d10ebd71b3f97da89fdac16f3e56","source":{"kind":"arxiv","id":"2605.13382","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13382","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13382v1","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13382","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"pith_short_12","alias_value":"55FVLLS257AN","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"55FVLLS257ANMDE2","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"55FVLLS2","created_at":"2026-05-18T12:33:37Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:55FVLLS257ANMDE2F5DU2B3ESM","target":"record","payload":{"canonical_record":{"source":{"id":"2605.13382","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2026-05-13T11:37:51Z","cross_cats_sorted":[],"title_canon_sha256":"b117477d7afc8cf4c8c6746c53e675be06c5f931688f7bea381ae57838da0a66","abstract_canon_sha256":"ecb922ceebb7025b32744b266b68f0d834cf5fc491a3a208020e76698e38da41"},"schema_version":"1.0"},"canonical_sha256":"ef4b55ae5aefc0d60c9a2f474d0764930420d10ebd71b3f97da89fdac16f3e56","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-18T02:44:47.824536Z","signature_b64":"LWx+PPd5hYf77QJG86gkmIZV5RIDowOB/cbIq9hlXNgGeCepThO1B+YdzFHgyhUW6oxPnmV2uPs+0KEop45DAg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"ef4b55ae5aefc0d60c9a2f474d0764930420d10ebd71b3f97da89fdac16f3e56","last_reissued_at":"2026-05-18T02:44:47.824098Z","signature_status":"signed_v1","first_computed_at":"2026-05-18T02:44:47.824098Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2605.13382","source_version":1,"attestation_state":"computed"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T02:44:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"5sbDIEU07nAprQ5528a84oXlbPFUAYQYClfl7SPoowrGK/d/dAUv0ZhRBUMYXg+VUX5/q0mgNMI5Dy9eMC9aAw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-26T17:44:55.643310Z"},"content_sha256":"b1236481b137aa667ede1b76aa1162459c87d294f39b7fa5bcc0fb5f2336ba71","schema_version":"1.0","event_id":"sha256:b1236481b137aa667ede1b76aa1162459c87d294f39b7fa5bcc0fb5f2336ba71"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:55FVLLS257ANMDE2F5DU2B3ESM","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"BlockVLA: Accelerating Autoregressive VLA via Block Diffusion Finetuning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"BlockVLA accelerates autoregressive VLA models by 3.3x using block diffusion finetuning, with faster training convergence and better early performance on long-horizon robotic tasks.","cross_cats":[],"primary_cat":"cs.RO","authors_text":"Badong Chen, Haoran Zhang, Ruiheng Wang, Shuanghao Bai, Xiangyu Xu","submitted_at":"2026-05-13T11:37:51Z","abstract_excerpt":"While autoregressive (AR) Vision-Language-Action (VLA) models have demonstrated formidable reasoning capabilities in robotic tasks, their sequential decoding process often incurs high inference latency and may amplify error accumulation during long-horizon execution. Discrete Diffusion Language Models (dLLMs) provide a promising alternative through parallel token refinement, but their practical deployment in robotics remains limited by repeated denoising function evaluations (NFEs) and the difficulty of directly applying standard KV caching to bidirectional iterative decoding. To bridge these "},"claims":{"count":3,"items":[{"kind":"strongest_claim","text":"BlockVLA achieves a 3.3× inference acceleration over standard discrete diffusion baselines and exhibits superior training efficiency with significant performance gains in the early stages of training on complex, long-horizon tasks.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That maintaining autoregressive dependencies only at the block level while performing parallel denoising inside blocks preserves the original model's reasoning capabilities and does not introduce new modes of error accumulation during long-horizon execution.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"BlockVLA accelerates autoregressive VLA models by 3.3x using block diffusion finetuning, with faster training convergence and better early performance on long-horizon robotic tasks.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"}],"snapshot_sha256":"a7c39b1d691aaaac853dff29ba9bfc075ffdc86f1c84b51674fb37ead2671d59"},"source":{"id":"2605.13382","kind":"arxiv","version":1},"verdict":{"id":"e6b665cd-48bc-4b5b-ae73-cfdbed5406c5","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T17:49:16.476101Z","strongest_claim":"BlockVLA achieves a 3.3× inference acceleration over standard discrete diffusion baselines and exhibits superior training efficiency with significant performance gains in the early stages of training on complex, long-horizon tasks.","one_line_summary":"BlockVLA accelerates autoregressive VLA models by 3.3x using block diffusion finetuning, with faster training convergence and better early performance on long-horizon robotic tasks.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That maintaining autoregressive dependencies only at the block level while performing parallel denoising inside blocks preserves the original model's reasoning capabilities and does not introduce new modes of error accumulation during long-horizon execution.","pith_extraction_headline":""},"references":{"count":20,"sample":[{"doi":"","year":null,"title":"OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Models","work_id":"87bfa84a-e663-4165-806f-93ef439d88d0","ref_index":1,"cited_arxiv_id":"2308.01390","is_internal_anchor":true},{"doi":"","year":null,"title":"Qwen Technical Report","work_id":"bb1fd52f-6b2f-437c-9516-37bdf6eb9be8","ref_index":2,"cited_arxiv_id":"2309.16609","is_internal_anchor":true},{"doi":"","year":null,"title":"Embodied robot manipulation in the era of foundation models: Planning and learning perspectives","work_id":"cfccbcb8-e5e5-462d-9ad0-ad7a780d7b94","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"Latent Reasoning VLA: Latent Thinking and Prediction for Vision-Language-Action Models","work_id":"6e132164-e81e-47c1-a2b0-0abae63e122f","ref_index":4,"cited_arxiv_id":"2602.01166","is_internal_anchor":true},{"doi":"","year":null,"title":"LLaDA2.0: Scaling Up Diffusion Language Models to 100B","work_id":"a1b1080d-0a91-44a4-8f70-2bf3e7a27e0b","ref_index":5,"cited_arxiv_id":"2512.15745","is_internal_anchor":true}],"resolved_work":20,"snapshot_sha256":"ab1f4aa19379c6048884ea2b10e41e26e4db7db10aa5ba484e2dac7fbbc34632","internal_anchors":11},"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"},"verdict_id":"e6b665cd-48bc-4b5b-ae73-cfdbed5406c5"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-05-18T02:44:47Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"z/HTY/bQ7C+wcTLUteBqZeuG0SKABMEWx1y0NvmOShNlhGnUWhzkavwnIEukyxDFnf3u1esWM8tRlhyw1mvgCw==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-05-26T17:44:55.644241Z"},"content_sha256":"7c7a05783ed6e7f666ee015aaa9d03450b6ea4464ee442c67c92d903260e3191","schema_version":"1.0","event_id":"sha256:7c7a05783ed6e7f666ee015aaa9d03450b6ea4464ee442c67c92d903260e3191"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/55FVLLS257ANMDE2F5DU2B3ESM/bundle.json","state_url":"https://pith.science/pith/55FVLLS257ANMDE2F5DU2B3ESM/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/55FVLLS257ANMDE2F5DU2B3ESM/bundle.json","status":"primary"}],"public_keys":[{"key_id":"pith-v1-2026-05","algorithm":"ed25519","format":"raw","public_key_b64":"stVStoiQhXFxp4s2pdzPNoqVNBMojDU/fJ2db5S3CbM=","public_key_hex":"b2d552b68890857171a78b36a5dccf368a953413288c353f7c9d9d6f94b709b3","fingerprint_sha256_b32_first128bits":"RVFV5Z2OI2J3ZUO7ERDEBCYNKS","fingerprint_sha256_hex":"8d4b5ee74e4693bcd1df2446408b0d54","rotates_at":null,"url":"https://pith.science/pith-signing-key.json","notes":"Pith uses this Ed25519 key to sign canonical record SHA-256 digests. Verify with: ed25519_verify(public_key, message=canonical_sha256_bytes, signature=base64decode(signature_b64))."}],"merge_version":"pith-open-graph-merge-v1","built_at":"2026-05-26T17:44:55Z","links":{"resolver":"https://pith.science/pith/55FVLLS257ANMDE2F5DU2B3ESM","bundle":"https://pith.science/pith/55FVLLS257ANMDE2F5DU2B3ESM/bundle.json","state":"https://pith.science/pith/55FVLLS257ANMDE2F5DU2B3ESM/state.json","well_known_bundle":"https://pith.science/.well-known/pith/55FVLLS257ANMDE2F5DU2B3ESM/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:55FVLLS257ANMDE2F5DU2B3ESM","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"ecb922ceebb7025b32744b266b68f0d834cf5fc491a3a208020e76698e38da41","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2026-05-13T11:37:51Z","title_canon_sha256":"b117477d7afc8cf4c8c6746c53e675be06c5f931688f7bea381ae57838da0a66"},"schema_version":"1.0","source":{"id":"2605.13382","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2605.13382","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"arxiv_version","alias_value":"2605.13382v1","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2605.13382","created_at":"2026-05-18T02:44:47Z"},{"alias_kind":"pith_short_12","alias_value":"55FVLLS257AN","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_16","alias_value":"55FVLLS257ANMDE2","created_at":"2026-05-18T12:33:37Z"},{"alias_kind":"pith_short_8","alias_value":"55FVLLS2","created_at":"2026-05-18T12:33:37Z"}],"graph_snapshots":[{"event_id":"sha256:7c7a05783ed6e7f666ee015aaa9d03450b6ea4464ee442c67c92d903260e3191","target":"graph","created_at":"2026-05-18T02:44:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"graph_snapshot":{"author_claims":{"count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","strong_count":0},"builder_version":"pith-number-builder-2026-05-17-v1","claims":{"count":3,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"BlockVLA achieves a 3.3× inference acceleration over standard discrete diffusion baselines and exhibits superior training efficiency with significant performance gains in the early stages of training on complex, long-horizon tasks."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That maintaining autoregressive dependencies only at the block level while performing parallel denoising inside blocks preserves the original model's reasoning capabilities and does not introduce new modes of error accumulation during long-horizon execution."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"BlockVLA accelerates autoregressive VLA models by 3.3x using block diffusion finetuning, with faster training convergence and better early performance on long-horizon robotic tasks."}],"snapshot_sha256":"a7c39b1d691aaaac853dff29ba9bfc075ffdc86f1c84b51674fb37ead2671d59"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"paper":{"abstract_excerpt":"While autoregressive (AR) Vision-Language-Action (VLA) models have demonstrated formidable reasoning capabilities in robotic tasks, their sequential decoding process often incurs high inference latency and may amplify error accumulation during long-horizon execution. Discrete Diffusion Language Models (dLLMs) provide a promising alternative through parallel token refinement, but their practical deployment in robotics remains limited by repeated denoising function evaluations (NFEs) and the difficulty of directly applying standard KV caching to bidirectional iterative decoding. To bridge these ","authors_text":"Badong Chen, Haoran Zhang, Ruiheng Wang, Shuanghao Bai, Xiangyu Xu","cross_cats":[],"headline":"BlockVLA accelerates autoregressive VLA models by 3.3x using block diffusion finetuning, with faster training convergence and better early performance on long-horizon robotic tasks.","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2026-05-13T11:37:51Z","title":"BlockVLA: Accelerating Autoregressive VLA via Block Diffusion Finetuning"},"references":{"count":20,"internal_anchors":11,"resolved_work":20,"sample":[{"cited_arxiv_id":"2308.01390","doi":"","is_internal_anchor":true,"ref_index":1,"title":"OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Models","work_id":"87bfa84a-e663-4165-806f-93ef439d88d0","year":null},{"cited_arxiv_id":"2309.16609","doi":"","is_internal_anchor":true,"ref_index":2,"title":"Qwen Technical Report","work_id":"bb1fd52f-6b2f-437c-9516-37bdf6eb9be8","year":null},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Embodied robot manipulation in the era of foundation models: Planning and learning perspectives","work_id":"cfccbcb8-e5e5-462d-9ad0-ad7a780d7b94","year":null},{"cited_arxiv_id":"2602.01166","doi":"","is_internal_anchor":true,"ref_index":4,"title":"Latent Reasoning VLA: Latent Thinking and Prediction for Vision-Language-Action Models","work_id":"6e132164-e81e-47c1-a2b0-0abae63e122f","year":null},{"cited_arxiv_id":"2512.15745","doi":"","is_internal_anchor":true,"ref_index":5,"title":"LLaDA2.0: Scaling Up Diffusion Language Models to 100B","work_id":"a1b1080d-0a91-44a4-8f70-2bf3e7a27e0b","year":null}],"snapshot_sha256":"ab1f4aa19379c6048884ea2b10e41e26e4db7db10aa5ba484e2dac7fbbc34632"},"source":{"id":"2605.13382","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-14T17:49:16.476101Z","id":"e6b665cd-48bc-4b5b-ae73-cfdbed5406c5","model_set":{"reader":"grok-4.3"},"one_line_summary":"BlockVLA accelerates autoregressive VLA models by 3.3x using block diffusion finetuning, with faster training convergence and better early performance on long-horizon robotic tasks.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"","strongest_claim":"BlockVLA achieves a 3.3× inference acceleration over standard discrete diffusion baselines and exhibits superior training efficiency with significant performance gains in the early stages of training on complex, long-horizon tasks.","weakest_assumption":"That maintaining autoregressive dependencies only at the block level while performing parallel denoising inside blocks preserves the original model's reasoning capabilities and does not introduce new modes of error accumulation during long-horizon execution."}},"verdict_id":"e6b665cd-48bc-4b5b-ae73-cfdbed5406c5"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:b1236481b137aa667ede1b76aa1162459c87d294f39b7fa5bcc0fb5f2336ba71","target":"record","created_at":"2026-05-18T02:44:47Z","signer":{"key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signer_id":"pith.science","signer_type":"pith_registry"},"payload":{"attestation_state":"computed","canonical_record":{"metadata":{"abstract_canon_sha256":"ecb922ceebb7025b32744b266b68f0d834cf5fc491a3a208020e76698e38da41","cross_cats_sorted":[],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.RO","submitted_at":"2026-05-13T11:37:51Z","title_canon_sha256":"b117477d7afc8cf4c8c6746c53e675be06c5f931688f7bea381ae57838da0a66"},"schema_version":"1.0","source":{"id":"2605.13382","kind":"arxiv","version":1}},"canonical_sha256":"ef4b55ae5aefc0d60c9a2f474d0764930420d10ebd71b3f97da89fdac16f3e56","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"ef4b55ae5aefc0d60c9a2f474d0764930420d10ebd71b3f97da89fdac16f3e56","first_computed_at":"2026-05-18T02:44:47.824098Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-18T02:44:47.824098Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"LWx+PPd5hYf77QJG86gkmIZV5RIDowOB/cbIq9hlXNgGeCepThO1B+YdzFHgyhUW6oxPnmV2uPs+0KEop45DAg==","signature_status":"signed_v1","signed_at":"2026-05-18T02:44:47.824536Z","signed_message":"canonical_sha256_bytes"},"source_id":"2605.13382","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:b1236481b137aa667ede1b76aa1162459c87d294f39b7fa5bcc0fb5f2336ba71","sha256:7c7a05783ed6e7f666ee015aaa9d03450b6ea4464ee442c67c92d903260e3191"],"state_sha256":"3c70f81efadce319585d905e3df21bf61db73b74230e9e8175882e8aff56ae99"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"oqC4RdaJUZQFzuFjvCCPd420lDf+NlS1ey5IdVsnGDDqiSUi0yyOQ5tD9/KHZxMFY5UkUvZQlWkFvVoPX0oNDw==","signed_message":"bundle_sha256_bytes","signed_at":"2026-05-26T17:44:55.648617Z","bundle_sha256":"2116311570a829ec27a8873ea41ea548ec9bb344b32e101f16f040a440ee7fa9"}}