{"bundle_type":"pith_open_graph_bundle","bundle_version":"1.0","pith_number":"pith:2026:NPKJA5VN3H4NDID6D744RVNIP4","short_pith_number":"pith:NPKJA5VN","canonical_record":{"source":{"id":"2608.03204","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-08-04T06:47:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"108058e5ee814b48eb5428a85666c30f8880ffc9106ae2da2f981c5f5d353286","abstract_canon_sha256":"a15a1370ec5982d255fbea0e0d0378f47ef31357945dd39ab0ebb72111ddfc6c"},"schema_version":"1.0"},"canonical_sha256":"6bd49076add9f8d1a07e1ff9c8d5a87f0d0fa546e0967f9aed7bd99e06996f7e","source":{"kind":"arxiv","id":"2608.03204","version":1},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.03204","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"arxiv_version","alias_value":"2608.03204v1","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.03204","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"pith_short_12","alias_value":"NPKJA5VN3H4N","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"pith_short_16","alias_value":"NPKJA5VN3H4NDID6","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"pith_short_8","alias_value":"NPKJA5VN","created_at":"2026-08-05T00:46:09Z"}],"events":[{"event_type":"record_created","subject_pith_number":"pith:2026:NPKJA5VN3H4NDID6D744RVNIP4","target":"record","payload":{"canonical_record":{"source":{"id":"2608.03204","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-08-04T06:47:08Z","cross_cats_sorted":["cs.AI"],"title_canon_sha256":"108058e5ee814b48eb5428a85666c30f8880ffc9106ae2da2f981c5f5d353286","abstract_canon_sha256":"a15a1370ec5982d255fbea0e0d0378f47ef31357945dd39ab0ebb72111ddfc6c"},"schema_version":"1.0"},"canonical_sha256":"6bd49076add9f8d1a07e1ff9c8d5a87f0d0fa546e0967f9aed7bd99e06996f7e","receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-08-05T00:46:09.931805Z","signature_b64":"+pBoHRFegjzWT9KPnAwQLUjOS+Dj63A9ochfkOcVeOqkoFLJdVxXsb70lqJkHO0V/+5Ac+IZq8f+R87/GNC/AA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"6bd49076add9f8d1a07e1ff9c8d5a87f0d0fa546e0967f9aed7bd99e06996f7e","last_reissued_at":"2026-08-05T00:46:09.929165Z","signature_status":"signed_v1","first_computed_at":"2026-08-05T00:46:09.929165Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"source_kind":"arxiv","source_id":"2608.03204","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":"2026-08-05T00:46:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"sfHjWbHLNbhCbkuyZeiJQMIGhzzqX6rQy/BQicZ6pRoNitfiEOUuz6St/txery6QOzTHeIAWBD30K2uI7IG9BQ==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T17:14:18.440979Z"},"content_sha256":"83b9c6fcdafb926c9e5df3d656aee43c5fd4af0244c99a0f009d83121d53420e","schema_version":"1.0","event_id":"sha256:83b9c6fcdafb926c9e5df3d656aee43c5fd4af0244c99a0f009d83121d53420e"},{"event_type":"graph_snapshot","subject_pith_number":"pith:2026:NPKJA5VN3H4NDID6D744RVNIP4","target":"graph","payload":{"graph_snapshot":{"paper":{"title":"Aligning Large Vision-Language Models at Test Time: A Trajectory-Guided Structured Sampling Approach","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.AI"],"primary_cat":"cs.CL","authors_text":"Gerard de Melo, Linlin Wang, Tianbao Jiang, Weicong Ni","submitted_at":"2026-08-04T06:47:08Z","abstract_excerpt":"Post-training reinforcement learning (RL) algorithms are commonly used to align large vision-language models (LVLMs) with human intent and the requirements of visual reasoning tasks. However, existing RL-based alignment methods are often resource-intensive and encounter mismatches between training objectives and inference-time distributions. To bridge this gap, we propose a novel test-time alignment approach that leverages trajectory-guided structured sampling for dynamic inference-time refinement, achieving better alignment with visual grounding and ensuring logical consistency. Our approach "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.03204","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/2608.03204/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"},"verdict_id":null},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-08-05T00:46:09Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"lQtahegtkSESrJTS22BDvbbWU6RABWNsTFjeKxm5npHFaS45a3SlDd77X6xcvkyJqmonEIUZn3WPVGdlVkx3Cg==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T17:14:18.441534Z"},"content_sha256":"23af0fbd51ab460d19b391d80d4ed161fa8ba8e8aab88031b91a70c7a5df6ff0","schema_version":"1.0","event_id":"sha256:23af0fbd51ab460d19b391d80d4ed161fa8ba8e8aab88031b91a70c7a5df6ff0"},{"event_type":"integrity_finding","subject_pith_number":"pith:2026:NPKJA5VN3H4NDID6D744RVNIP4","target":"integrity","payload":{"note":"DOI in the printed bibliography is fragmented by whitespace or line breaks. A longer candidate (10.18653/v1/2025.emnlp-main.922) was visible in the surrounding text but could not be confirmed against doi.org as printed.","snippet":"Ming Li, Nan Zhang, Chenrui Fan, Hong Jiao, Yanbin Fu, Sydney Peters, Qingshu Xu, Robert Lissitz, and Tianyi Zhou. 2025. Understanding the Thinking Pro- cess of Reasoning Models: A Perspective from Schoenfeld’s Episode Theory. In Proceeding","arxiv_id":"2608.03204","detector":"doi_compliance","evidence":{"ref_index":17,"verdict_class":"incontrovertible","resolved_title":null,"printed_excerpt":"10.18653/v1/2025.emn","reconstructed_doi":"10.18653/v1/2025.emnlp-main.922"},"severity":"advisory","ref_index":17,"audited_at":"2026-08-05T23:48:19.700364Z","event_type":"pith.integrity.v1","detected_doi":"10.18653/v1/2025.emnlp-main.922","detector_url":"https://pith.science/pith-integrity-protocol#doi_compliance","external_url":null,"finding_type":"recoverable_identifier","evidence_hash":"48cada5412f4d9fededb590711a59bbd6b0718ff8c588e105968e64a6101df85","paper_version":1,"verdict_class":"incontrovertible","resolved_title":null,"detector_version":"1.1.0","detected_arxiv_id":null,"integrity_event_id":18166,"payload_sha256":"28e932dcdfdaab0be35eb2082d7dc5bfa5ec6e724eccb57133bdc25a01b46489","signature_b64":"/xAQSATvmhtMn2GfUlkjIJoaSPiPs2RxZ2UD4HvtkhADgcizWmiB6vGbT3q6N4xNUKrPW/S8ot65an9U/m9eBg==","signing_key_id":"pith-v1-2026-05"},"signer":{"signer_id":"pith.science","signer_type":"pith_registry","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"created_at":"2026-08-05T23:48:28Z","supersedes":[],"prev_event":null,"signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"TVcCGhxh3+B2iVvv9c6M80Fq5OVz/poawibAsdSj+MB7CMhsil4a873WccJRErSxe5Y6O6hl3BPkkE+orXV7CA==","signed_message":"open_graph_event_sha256_bytes","signed_at":"2026-08-07T17:14:18.446839Z"},"content_sha256":"652b73de6e86a9ba1cdf14d482c3566abf2c0e61eac366edd560a3bcc031936c","schema_version":"1.0","event_id":"sha256:652b73de6e86a9ba1cdf14d482c3566abf2c0e61eac366edd560a3bcc031936c"}],"timestamp_proofs":[],"mirror_hints":[{"mirror_type":"https","name":"Pith Resolver","base_url":"https://pith.science","bundle_url":"https://pith.science/pith/NPKJA5VN3H4NDID6D744RVNIP4/bundle.json","state_url":"https://pith.science/pith/NPKJA5VN3H4NDID6D744RVNIP4/state.json","well_known_bundle_url":"https://pith.science/.well-known/pith/NPKJA5VN3H4NDID6D744RVNIP4/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":"2026-08-07T17:14:18Z","links":{"resolver":"https://pith.science/pith/NPKJA5VN3H4NDID6D744RVNIP4","bundle":"https://pith.science/pith/NPKJA5VN3H4NDID6D744RVNIP4/bundle.json","state":"https://pith.science/pith/NPKJA5VN3H4NDID6D744RVNIP4/state.json","well_known_bundle":"https://pith.science/.well-known/pith/NPKJA5VN3H4NDID6D744RVNIP4/bundle.json"},"state":{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2026:NPKJA5VN3H4NDID6D744RVNIP4","merge_version":"pith-open-graph-merge-v1","event_count":3,"valid_event_count":3,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"a15a1370ec5982d255fbea0e0d0378f47ef31357945dd39ab0ebb72111ddfc6c","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-08-04T06:47:08Z","title_canon_sha256":"108058e5ee814b48eb5428a85666c30f8880ffc9106ae2da2f981c5f5d353286"},"schema_version":"1.0","source":{"id":"2608.03204","kind":"arxiv","version":1}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2608.03204","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"arxiv_version","alias_value":"2608.03204v1","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2608.03204","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"pith_short_12","alias_value":"NPKJA5VN3H4N","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"pith_short_16","alias_value":"NPKJA5VN3H4NDID6","created_at":"2026-08-05T00:46:09Z"},{"alias_kind":"pith_short_8","alias_value":"NPKJA5VN","created_at":"2026-08-05T00:46:09Z"}],"graph_snapshots":[{"event_id":"sha256:23af0fbd51ab460d19b391d80d4ed161fa8ba8e8aab88031b91a70c7a5df6ff0","target":"graph","created_at":"2026-08-05T00:46:09Z","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":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"integrity":{"available":true,"clean":true,"detectors_run":[],"endpoint":"/pith/2608.03204/integrity.json","findings":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938","summary":{"advisory":0,"by_detector":{},"critical":0,"informational":0}},"paper":{"abstract_excerpt":"Post-training reinforcement learning (RL) algorithms are commonly used to align large vision-language models (LVLMs) with human intent and the requirements of visual reasoning tasks. However, existing RL-based alignment methods are often resource-intensive and encounter mismatches between training objectives and inference-time distributions. To bridge this gap, we propose a novel test-time alignment approach that leverages trajectory-guided structured sampling for dynamic inference-time refinement, achieving better alignment with visual grounding and ensuring logical consistency. Our approach ","authors_text":"Gerard de Melo, Linlin Wang, Tianbao Jiang, Weicong Ni","cross_cats":["cs.AI"],"headline":"","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-08-04T06:47:08Z","title":"Aligning Large Vision-Language Models at Test Time: A Trajectory-Guided Structured Sampling Approach"},"references":{"count":0,"internal_anchors":0,"resolved_work":0,"sample":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2608.03204","kind":"arxiv","version":1},"verdict":{"created_at":null,"id":null,"model_set":{},"one_line_summary":"","pipeline_version":null,"pith_extraction_headline":"","strongest_claim":"","weakest_assumption":""}},"verdict_id":null}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:83b9c6fcdafb926c9e5df3d656aee43c5fd4af0244c99a0f009d83121d53420e","target":"record","created_at":"2026-08-05T00:46:09Z","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":"a15a1370ec5982d255fbea0e0d0378f47ef31357945dd39ab0ebb72111ddfc6c","cross_cats_sorted":["cs.AI"],"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CL","submitted_at":"2026-08-04T06:47:08Z","title_canon_sha256":"108058e5ee814b48eb5428a85666c30f8880ffc9106ae2da2f981c5f5d353286"},"schema_version":"1.0","source":{"id":"2608.03204","kind":"arxiv","version":1}},"canonical_sha256":"6bd49076add9f8d1a07e1ff9c8d5a87f0d0fa546e0967f9aed7bd99e06996f7e","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"6bd49076add9f8d1a07e1ff9c8d5a87f0d0fa546e0967f9aed7bd99e06996f7e","first_computed_at":"2026-08-05T00:46:09.929165Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-08-05T00:46:09.929165Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"+pBoHRFegjzWT9KPnAwQLUjOS+Dj63A9ochfkOcVeOqkoFLJdVxXsb70lqJkHO0V/+5Ac+IZq8f+R87/GNC/AA==","signature_status":"signed_v1","signed_at":"2026-08-05T00:46:09.931805Z","signed_message":"canonical_sha256_bytes"},"source_id":"2608.03204","source_kind":"arxiv","source_version":1}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:83b9c6fcdafb926c9e5df3d656aee43c5fd4af0244c99a0f009d83121d53420e","sha256:23af0fbd51ab460d19b391d80d4ed161fa8ba8e8aab88031b91a70c7a5df6ff0","sha256:652b73de6e86a9ba1cdf14d482c3566abf2c0e61eac366edd560a3bcc031936c"],"state_sha256":"d0ac4a5dcb723cd00824c9df5d216f55804bdb21db7e1c4f6f9b92da0cbac9c5"},"bundle_signature":{"signature_status":"signed_v1","algorithm":"ed25519","key_id":"pith-v1-2026-05","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","signature_b64":"wr8dg2va1RSpv2IyqjwURoVMJhw0rV/ATUVFcWnAHyUZGEgClsu1+j5yCWWRqiSOUchpxxUV/rm/jiqNhwH0Ag==","signed_message":"bundle_sha256_bytes","signed_at":"2026-08-07T17:14:18.455619Z","bundle_sha256":"07c69dd4297a1a575b38e5237ce6242f58e0d64cc7e8c644fdd50a5771d8d4fc"}}