{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2025:QZ2JLTFX3WVH7R5YJ3DKA7WYBR","short_pith_number":"pith:QZ2JLTFX","canonical_record":{"source":{"id":"2501.09747","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-01-16T18:57:04Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"813f6131b478b8a82b4648da930ab45df305324752fb2beb9d52ef93948d5491","abstract_canon_sha256":"ee00ef47fddef2e54671930a321ff3a362436539b53f0bfae5bc443e7df83f20"},"schema_version":"1.0"},"canonical_sha256":"867495ccb7ddaa7fc7b84ec6a07ed80c73b6d783e8bed27c577a9b6402bcf3ec","source":{"kind":"arxiv","id":"2501.09747","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.09747","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"arxiv_version","alias_value":"2501.09747v1","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.09747","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"pith_short_12","alias_value":"QZ2JLTFX3WVH","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"pith_short_16","alias_value":"QZ2JLTFX3WVH7R5Y","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"pith_short_8","alias_value":"QZ2JLTFX","created_at":"2026-07-05T10:01:58Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2025:QZ2JLTFX3WVH7R5YJ3DKA7WYBR","target":"record","payload":{"canonical_record":{"source":{"id":"2501.09747","kind":"arxiv","version":1},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-01-16T18:57:04Z","cross_cats_sorted":["cs.LG"],"title_canon_sha256":"813f6131b478b8a82b4648da930ab45df305324752fb2beb9d52ef93948d5491","abstract_canon_sha256":"ee00ef47fddef2e54671930a321ff3a362436539b53f0bfae5bc443e7df83f20"},"schema_version":"1.0"},"canonical_sha256":"867495ccb7ddaa7fc7b84ec6a07ed80c73b6d783e8bed27c577a9b6402bcf3ec","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:01:58.002105Z","signature_b64":"+i8DLvpspPSZJnzzifFyszIIc6ReuVwfnc3my+TXmld6JWIqsomTs4inaXgKZmQ534NHDbSZ0j2MXoq/wTGCDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"867495ccb7ddaa7fc7b84ec6a07ed80c73b6d783e8bed27c577a9b6402bcf3ec","last_reissued_at":"2026-07-05T10:01:58.001618Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:01:58.001618Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2501.09747","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":"2026-07-05T10:01:58Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"PnAOTl4YN39rip0blr8MUEpGFcb0n7SAPc/j2Ou3SBvZWys+4hmWMKh66+4jzEhiZ5r9vW4UpCM43YgmKTxvBA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T05:03:49.046844Z"},"content_sha256":"8d8ae0b68e905aaf28bffe3dde5ed9a188923cf86e564c002c1ec0ab12408450","schema_version":"1.0","event_id":"sha256:8d8ae0b68e905aaf28bffe3dde5ed9a188923cf86e564c002c1ec0ab12408450"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2025:QZ2JLTFX3WVH7R5YJ3DKA7WYBR","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"FAST: Efficient Action Tokenization for Vision-Language-Action Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Frequency-space tokenization allows autoregressive VLAs to succeed on dexterous high-frequency robot tasks where standard binning fails.","cross_cats":["cs.LG"],"primary_cat":"cs.RO","authors_text":"Brian Ichter, Chelsea Finn, Danny Driess, Karl Pertsch, Kyle Stachowicz, Oier Mees, Quan Vuong, Sergey Levine, Suraj Nair","submitted_at":"2025-01-16T18:57:04Z","abstract_excerpt":"Autoregressive sequence models, such as Transformer-based vision-language action (VLA) policies, can be tremendously effective for capturing complex and generalizable robotic behaviors. However, such models require us to choose a tokenization of our continuous action signals, which determines how the discrete symbols predicted by the model map to continuous robot actions. We find that current approaches for robot action tokenization, based on simple per-dimension, per-timestep binning schemes, typically perform poorly when learning dexterous skills from high-frequency robot data. To address th"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our tokenization approach, Frequency-space Action Sequence Tokenization (FAST), enables us to train autoregressive VLAs for highly dexterous and high-frequency tasks where standard discretization methods fail completely. Based on FAST, we release FAST+, a universal robot action tokenizer, trained on 1M real robot action trajectories. Finally, we show that, when combined with the pi0 VLA, our method can scale to training on 10k hours of robot data and match the performance of diffusion VLAs, while reducing training time by up to 5x.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That frequency-domain compression via the discrete cosine transform preserves all information necessary for precise, stable closed-loop control in dexterous high-frequency tasks without introducing artifacts that standard binning avoids.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"FAST applies discrete cosine transform to robot action sequences for efficient tokenization, enabling autoregressive VLAs to succeed on high-frequency dexterous tasks and scale to 10k hours of data while matching diffusion VLA performance with up to 5x faster training.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Frequency-space tokenization allows autoregressive VLAs to succeed on dexterous high-frequency robot tasks where standard binning fails.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"bf6035bdc590a94ecf6a843f3798e49ab8d25cf12de73662430890b0ac150a1e"},"source":{"id":"2501.09747","kind":"arxiv","version":1},"verdict":{"id":"f7efb8ed-dfa7-413e-86ea-17df665ed269","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-11T08:46:46.481344Z","strongest_claim":"Our tokenization approach, Frequency-space Action Sequence Tokenization (FAST), enables us to train autoregressive VLAs for highly dexterous and high-frequency tasks where standard discretization methods fail completely. Based on FAST, we release FAST+, a universal robot action tokenizer, trained on 1M real robot action trajectories. Finally, we show that, when combined with the pi0 VLA, our method can scale to training on 10k hours of robot data and match the performance of diffusion VLAs, while reducing training time by up to 5x.","one_line_summary":"FAST applies discrete cosine transform to robot action sequences for efficient tokenization, enabling autoregressive VLAs to succeed on high-frequency dexterous tasks and scale to 10k hours of data while matching diffusion VLA performance with up to 5x faster training.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That frequency-domain compression via the discrete cosine transform preserves all information necessary for precise, stable closed-loop control in dexterous high-frequency tasks without introducing artifacts that standard binning avoids.","pith_extraction_headline":"Frequency-space tokenization allows autoregressive VLAs to succeed on dexterous high-frequency robot tasks where standard binning fails."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2501.09747/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":73,"sample":[{"doi":"","year":1974,"title":"Dis- crete cosine transform","work_id":"d0188da4-50a2-4b27-92e8-441b9590c285","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Do As I Can, Not As I Say: Grounding Language in Robotic Affordances","work_id":"037320f1-b0a9-4cbe-a639-bfb25409ce71","ref_index":2,"cited_arxiv_id":"2204.01691","is_internal_anchor":true},{"doi":"","year":2024,"title":"Minivla: A better vla with a smaller footprint","work_id":"5feb606f-b7fc-495a-ae0e-89240411e3fd","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2024,"title":"RT-H: Action Hierarchies Using Language","work_id":"ecf7cf18-c1a8-4a6b-bc2a-fb165643aa0d","ref_index":4,"cited_arxiv_id":"2403.01823","is_internal_anchor":true},{"doi":"","year":2024,"title":"PaliGemma: A versatile 3B VLM for transfer","work_id":"df6f48b3-5792-47c7-9614-cb856ea31ad9","ref_index":5,"cited_arxiv_id":"2407.07726","is_internal_anchor":true}],"resolved_work":73,"snapshot_sha256":"c1c7e601245d4af16a38005f9ef7623f3b989448cfb1307b52746e73c2a88e88","internal_anchors":18},"formal_canon":{"evidence_count":2,"snapshot_sha256":"9a6bd1066c462aab8615ecdc9d308d808ba9fea5550e5d1d1c719a7624881194"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"verdict_id":"f7efb8ed-dfa7-413e-86ea-17df665ed269"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-07-05T10:01:58Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"0pLdBWbxwZ3Maw+kqjMzevCI60Cn/oDEvAKN+fZ/cGm8ZPOsEojuN3CtbUaxdfq6kxHnRzSAUGgYNmW2g6taDg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-06T05:03:49.047801Z"},"content_sha256":"2a712631e1c413396dd605653747668517fc815f288b7a15c7566cab0ceac82f","schema_version":"1.0","event_id":"sha256:2a712631e1c413396dd605653747668517fc815f288b7a15c7566cab0ceac82f"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/QZ2JLTFX3WVH7R5YJ3DKA7WYBR/bundle.json","state_url":"https://pith.science/pith/QZ2JLTFX3WVH7R5YJ3DKA7WYBR/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/QZ2JLTFX3WVH7R5YJ3DKA7WYBR/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":"2026-08-06T05:03:49Z","links":{"resolver":"https://pith.science/pith/QZ2JLTFX3WVH7R5YJ3DKA7WYBR","bundle":"https://pith.science/pith/QZ2JLTFX3WVH7R5YJ3DKA7WYBR/bundle.json","state":"https://pith.science/pith/QZ2JLTFX3WVH7R5YJ3DKA7WYBR/state.json","well_known_bundle":"https://pith.science/.well-known/pith/QZ2JLTFX3WVH7R5YJ3DKA7WYBR/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2025:QZ2JLTFX3WVH7R5YJ3DKA7WYBR","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":"ee00ef47fddef2e54671930a321ff3a362436539b53f0bfae5bc443e7df83f20","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-01-16T18:57:04Z","title_canon_sha256":"813f6131b478b8a82b4648da930ab45df305324752fb2beb9d52ef93948d5491"},"schema_version":"1.0","source":{"id":"2501.09747","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2501.09747","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"arxiv_version","alias_value":"2501.09747v1","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.09747","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"pith_short_12","alias_value":"QZ2JLTFX3WVH","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"pith_short_16","alias_value":"QZ2JLTFX3WVH7R5Y","created_at":"2026-07-05T10:01:58Z"},{"alias_kind":"pith_short_8","alias_value":"QZ2JLTFX","created_at":"2026-07-05T10:01:58Z"}],"graph_snapshots":[{"event_id":"sha256:2a712631e1c413396dd605653747668517fc815f288b7a15c7566cab0ceac82f","target":"graph","created_at":"2026-07-05T10:01:58Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Our tokenization approach, Frequency-space Action Sequence Tokenization (FAST), enables us to train autoregressive VLAs for highly dexterous and high-frequency tasks where standard discretization methods fail completely. Based on FAST, we release FAST+, a universal robot action tokenizer, trained on 1M real robot action trajectories. Finally, we show that, when combined with the pi0 VLA, our method can scale to training on 10k hours of robot data and match the performance of diffusion VLAs, while reducing training time by up to 5x."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"That frequency-domain compression via the discrete cosine transform preserves all information necessary for precise, stable closed-loop control in dexterous high-frequency tasks without introducing artifacts that standard binning avoids."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"FAST applies discrete cosine transform to robot action sequences for efficient tokenization, enabling autoregressive VLAs to succeed on high-frequency dexterous tasks and scale to 10k hours of data while matching diffusion VLA performance with up to 5x faster training."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"Frequency-space tokenization allows autoregressive VLAs to succeed on dexterous high-frequency robot tasks where standard binning fails."}],"snapshot_sha256":"bf6035bdc590a94ecf6a843f3798e49ab8d25cf12de73662430890b0ac150a1e"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"9a6bd1066c462aab8615ecdc9d308d808ba9fea5550e5d1d1c719a7624881194"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2501.09747/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Autoregressive sequence models, such as Transformer-based vision-language action (VLA) policies, can be tremendously effective for capturing complex and generalizable robotic behaviors. However, such models require us to choose a tokenization of our continuous action signals, which determines how the discrete symbols predicted by the model map to continuous robot actions. We find that current approaches for robot action tokenization, based on simple per-dimension, per-timestep binning schemes, typically perform poorly when learning dexterous skills from high-frequency robot data. To address th","authors_text":"Brian Ichter, Chelsea Finn, Danny Driess, Karl Pertsch, Kyle Stachowicz, Oier Mees, Quan Vuong, Sergey Levine, Suraj Nair","cross_cats":["cs.LG"],"headline":"Frequency-space tokenization allows autoregressive VLAs to succeed on dexterous high-frequency robot tasks where standard binning fails.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-01-16T18:57:04Z","title":"FAST: Efficient Action Tokenization for Vision-Language-Action Models"},"references":{"count":73,"internal_anchors":18,"resolved_work":73,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"Dis- crete cosine transform","work_id":"d0188da4-50a2-4b27-92e8-441b9590c285","year":1974},{"cited_arxiv_id":"2204.01691","doi":"","is_internal_anchor":true,"ref_index":2,"title":"Do As I Can, Not As I Say: Grounding Language in Robotic Affordances","work_id":"037320f1-b0a9-4cbe-a639-bfb25409ce71","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Minivla: A better vla with a smaller footprint","work_id":"5feb606f-b7fc-495a-ae0e-89240411e3fd","year":2024},{"cited_arxiv_id":"2403.01823","doi":"","is_internal_anchor":true,"ref_index":4,"title":"RT-H: Action Hierarchies Using Language","work_id":"ecf7cf18-c1a8-4a6b-bc2a-fb165643aa0d","year":2024},{"cited_arxiv_id":"2407.07726","doi":"","is_internal_anchor":true,"ref_index":5,"title":"PaliGemma: A versatile 3B VLM for transfer","work_id":"df6f48b3-5792-47c7-9614-cb856ea31ad9","year":2024}],"snapshot_sha256":"c1c7e601245d4af16a38005f9ef7623f3b989448cfb1307b52746e73c2a88e88"},"source":{"id":"2501.09747","kind":"arxiv","version":1},"verdict":{"created_at":"2026-05-11T08:46:46.481344Z","id":"f7efb8ed-dfa7-413e-86ea-17df665ed269","model_set":{"reader":"grok-4.3"},"one_line_summary":"FAST applies discrete cosine transform to robot action sequences for efficient tokenization, enabling autoregressive VLAs to succeed on high-frequency dexterous tasks and scale to 10k hours of data while matching diffusion VLA performance with up to 5x faster training.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"Frequency-space tokenization allows autoregressive VLAs to succeed on dexterous high-frequency robot tasks where standard binning fails.","strongest_claim":"Our tokenization approach, Frequency-space Action Sequence Tokenization (FAST), enables us to train autoregressive VLAs for highly dexterous and high-frequency tasks where standard discretization methods fail completely. Based on FAST, we release FAST+, a universal robot action tokenizer, trained on 1M real robot action trajectories. Finally, we show that, when combined with the pi0 VLA, our method can scale to training on 10k hours of robot data and match the performance of diffusion VLAs, while reducing training time by up to 5x.","weakest_assumption":"That frequency-domain compression via the discrete cosine transform preserves all information necessary for precise, stable closed-loop control in dexterous high-frequency tasks without introducing artifacts that standard binning avoids."}},"verdict_id":"f7efb8ed-dfa7-413e-86ea-17df665ed269"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:8d8ae0b68e905aaf28bffe3dde5ed9a188923cf86e564c002c1ec0ab12408450","target":"record","created_at":"2026-07-05T10:01:58Z","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":"ee00ef47fddef2e54671930a321ff3a362436539b53f0bfae5bc443e7df83f20","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.RO","submitted_at":"2025-01-16T18:57:04Z","title_canon_sha256":"813f6131b478b8a82b4648da930ab45df305324752fb2beb9d52ef93948d5491"},"schema_version":"1.0","source":{"id":"2501.09747","kind":"arxiv","version":1}},"canonical_sha256":"867495ccb7ddaa7fc7b84ec6a07ed80c73b6d783e8bed27c577a9b6402bcf3ec","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"867495ccb7ddaa7fc7b84ec6a07ed80c73b6d783e8bed27c577a9b6402bcf3ec","first_computed_at":"2026-07-05T10:01:58.001618Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-07-05T10:01:58.001618Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"+i8DLvpspPSZJnzzifFyszIIc6ReuVwfnc3my+TXmld6JWIqsomTs4inaXgKZmQ534NHDbSZ0j2MXoq/wTGCDw==","signature_status":"signed_v1","signed_at":"2026-07-05T10:01:58.002105Z","signed_message":"canonical_sha256_bytes"},"source_id":"2501.09747","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:8d8ae0b68e905aaf28bffe3dde5ed9a188923cf86e564c002c1ec0ab12408450","sha256:2a712631e1c413396dd605653747668517fc815f288b7a15c7566cab0ceac82f"],"state_sha256":"a7774e4ecd00cdd17ab56c5b1940de53a8fabda58c8c43ff1bf2782715fddd49"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"6mRmjsm4XUdO2tlB+LlLbC5Mr+Q5j35q0O/OuRR6INEND4BdM/WaNPNCbuKINVVfuDHqREKS96izxTRNTTG6Ag==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-06T05:03:49.053130Z","bundle_sha256":"5ace93eee33d3918e9b6f28bfb7371d0584e8d867d67b27ccdb8d16388115fb4"}}