{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:CRLQZXOY55NK45P7UY3HL7K2MQ","short_pith_number":"pith:CRLQZXOY","schema_version":"1.0","canonical_sha256":"14570cddd8ef5aae75ffa63675fd5a640f32af29c32fbfce12e15608d1288c47","source":{"kind":"arxiv","id":"2503.06514","version":2},"attestation_state":"computed","paper":{"title":"GFlowVLM: Enhancing Multi-step Reasoning in Vision-Language Models with Generative Flow Networks","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CL","authors_text":"Enna Sachdeva, Haoqiang Kang, Kwonjoon Lee, Piyush Gupta, Sangjae Bae","submitted_at":"2025-03-09T08:38:10Z","abstract_excerpt":"Vision-Language Models (VLMs) have recently shown promising advancements in sequential decision-making tasks through task-specific fine-tuning. However, common fine-tuning methods, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) techniques like Proximal Policy Optimization (PPO), present notable limitations: SFT assumes Independent and Identically Distributed (IID) data, while PPO focuses on maximizing cumulative rewards. These limitations often restrict solution diversity and hinder generalization in multi-step reasoning tasks. To address these challenges, we introduce a "},"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":"2503.06514","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","primary_cat":"cs.CL","submitted_at":"2025-03-09T08:38:10Z","cross_cats_sorted":["cs.AI","cs.CV","cs.LG"],"title_canon_sha256":"edbc5e4fc0a603a3f2b58f66a1f4113d7a02edcf492255d7fbdce998ac024680","abstract_canon_sha256":"ec28f303e51ce7ec91941df0dfb4ea7ce3f470e3e32351c1fcfd566df1de4bca"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T10:38:44.007332Z","signature_b64":"wdJh1OJMBaVCta3fApH4C7jiSPbt/SKQO51yL4WfVHdgbPkW8xX0tEDEbm6WTtgMAYfAEZXYFt/eTuzieGakAQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"14570cddd8ef5aae75ffa63675fd5a640f32af29c32fbfce12e15608d1288c47","last_reissued_at":"2026-07-05T10:38:44.006861Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T10:38:44.006861Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"GFlowVLM: Enhancing Multi-step Reasoning in Vision-Language Models with Generative Flow Networks","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"","cross_cats":["cs.AI","cs.CV","cs.LG"],"primary_cat":"cs.CL","authors_text":"Enna Sachdeva, Haoqiang Kang, Kwonjoon Lee, Piyush Gupta, Sangjae Bae","submitted_at":"2025-03-09T08:38:10Z","abstract_excerpt":"Vision-Language Models (VLMs) have recently shown promising advancements in sequential decision-making tasks through task-specific fine-tuning. However, common fine-tuning methods, such as Supervised Fine-Tuning (SFT) and Reinforcement Learning (RL) techniques like Proximal Policy Optimization (PPO), present notable limitations: SFT assumes Independent and Identically Distributed (IID) data, while PPO focuses on maximizing cumulative rewards. These limitations often restrict solution diversity and hinder generalization in multi-step reasoning tasks. To address these challenges, we introduce a "},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2503.06514","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/2503.06514/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":"2503.06514","created_at":"2026-07-05T10:38:44.006917+00:00"},{"alias_kind":"arxiv_version","alias_value":"2503.06514v2","created_at":"2026-07-05T10:38:44.006917+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2503.06514","created_at":"2026-07-05T10:38:44.006917+00:00"},{"alias_kind":"pith_short_12","alias_value":"CRLQZXOY55NK","created_at":"2026-07-05T10:38:44.006917+00:00"},{"alias_kind":"pith_short_16","alias_value":"CRLQZXOY55NK45P7","created_at":"2026-07-05T10:38:44.006917+00:00"},{"alias_kind":"pith_short_8","alias_value":"CRLQZXOY","created_at":"2026-07-05T10:38:44.006917+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":1,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2506.16114","citing_title":"GFlowGR: Fine-tuning Generative Recommendation Frameworks with Generative Flow Networks","ref_index":2025,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ","json":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ.json","graph_json":"https://pith.science/api/pith-number/CRLQZXOY55NK45P7UY3HL7K2MQ/graph.json","events_json":"https://pith.science/api/pith-number/CRLQZXOY55NK45P7UY3HL7K2MQ/events.json","paper":"https://pith.science/paper/CRLQZXOY"},"agent_actions":{"view_html":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ","download_json":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ.json","view_paper":"https://pith.science/paper/CRLQZXOY","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2503.06514&json=true","fetch_graph":"https://pith.science/api/pith-number/CRLQZXOY55NK45P7UY3HL7K2MQ/graph.json","fetch_events":"https://pith.science/api/pith-number/CRLQZXOY55NK45P7UY3HL7K2MQ/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ/action/timestamp_anchor","attest_storage":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ/action/storage_attestation","attest_author":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ/action/author_attestation","sign_citation":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ/action/citation_signature","submit_replication":"https://pith.science/pith/CRLQZXOY55NK45P7UY3HL7K2MQ/action/replication_record"}},"created_at":"2026-07-05T10:38:44.006917+00:00","updated_at":"2026-07-05T10:38:44.006917+00:00"}