{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:R6H35SOKFLENDXN3ADDGNKWUEO","short_pith_number":"pith:R6H35SOK","schema_version":"1.0","canonical_sha256":"8f8fbec9ca2ac8d1ddbb00c666aad423859da517782b1709711bf019f78f815c","source":{"kind":"arxiv","id":"2511.13649","version":5},"attestation_state":"computed","paper":{"title":"Distribution Matching Distillation Meets Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bo Zhang, Changsheng Lu, David Liu, Dengyang Jiang, Dongyang Liu, Harry Yang, Hengzhuang Li, Liuzhuozheng Li, Mengmeng Wang, Peng Gao, Qilong Wu, Steven Hoi, Xin Jin, Zanyi Wang, Zhen Li","submitted_at":"2025-11-17T17:59:54Z","abstract_excerpt":"Distribution Matching Distillation (DMD) facilitates efficient inference by distilling multi-step diffusion models into few-step variants. Concurrently, Reinforcement Learning (RL) has emerged as a vital tool for aligning generative models with human preferences. While both represent critical post-training stages for large-scale diffusion models, existing studies typically treat them as independent, sequential processes, leaving a systematic framework for their unification largely unexplored. In this work, we demonstrate that jointly optimizing these two objectives yields mutual benefits: RL e"},"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":"2511.13649","kind":"arxiv","version":5},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-11-17T17:59:54Z","cross_cats_sorted":[],"title_canon_sha256":"5d5aada1fba707d0785c07183b782f75bdf2441e4171fa91b663a1afc32912c1","abstract_canon_sha256":"01d1a7c06bef03a2cf7cb6d0642716c58fbf0c81705d5a3e479fbaceb3363eb5"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-07T02:17:13.730773Z","signature_b64":"AZN89UR9q+J0ctslnsZloVROnrrB2/2GXWwllX2RPQEi1TIj8wKINkboGShwl1vavRE2A+aQSJUvmxCl93H1Dg==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8f8fbec9ca2ac8d1ddbb00c666aad423859da517782b1709711bf019f78f815c","last_reissued_at":"2026-07-07T02:17:13.729789Z","signature_status":"signed_v1","first_computed_at":"2026-07-07T02:17:13.729789Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Distribution Matching Distillation Meets Reinforcement Learning","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bo Zhang, Changsheng Lu, David Liu, Dengyang Jiang, Dongyang Liu, Harry Yang, Hengzhuang Li, Liuzhuozheng Li, Mengmeng Wang, Peng Gao, Qilong Wu, Steven Hoi, Xin Jin, Zanyi Wang, Zhen Li","submitted_at":"2025-11-17T17:59:54Z","abstract_excerpt":"Distribution Matching Distillation (DMD) facilitates efficient inference by distilling multi-step diffusion models into few-step variants. Concurrently, Reinforcement Learning (RL) has emerged as a vital tool for aligning generative models with human preferences. While both represent critical post-training stages for large-scale diffusion models, existing studies typically treat them as independent, sequential processes, leaving a systematic framework for their unification largely unexplored. In this work, we demonstrate that jointly optimizing these two objectives yields mutual benefits: RL e"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2511.13649","kind":"arxiv","version":5},"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/2511.13649/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":"2511.13649","created_at":"2026-07-07T02:17:13.729907+00:00"},{"alias_kind":"arxiv_version","alias_value":"2511.13649v5","created_at":"2026-07-07T02:17:13.729907+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2511.13649","created_at":"2026-07-07T02:17:13.729907+00:00"},{"alias_kind":"pith_short_12","alias_value":"R6H35SOKFLEN","created_at":"2026-07-07T02:17:13.729907+00:00"},{"alias_kind":"pith_short_16","alias_value":"R6H35SOKFLENDXN3","created_at":"2026-07-07T02:17:13.729907+00:00"},{"alias_kind":"pith_short_8","alias_value":"R6H35SOK","created_at":"2026-07-07T02:17:13.729907+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":28,"internal_anchor_count":28,"sample":[{"citing_arxiv_id":"2606.25473","citing_title":"Causal-rCM: A Unified Teacher-Forcing and Self-Forcing Open Recipe for Autoregressive Diffusion Distillation in Streaming Video Generation and Interactive World Models","ref_index":28,"is_internal_anchor":true},{"citing_arxiv_id":"2606.21982","citing_title":"CoDMD: Copula-aware Distribution Matching Distillation for Fast Video Generation","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2606.21304","citing_title":"A Test-time Actor-Critic Approach to News Images Generation","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2606.12575","citing_title":"High-Fidelity Two-Step Image Generation via Teacher-Aligned End-to-End Distillation","ref_index":9,"is_internal_anchor":true},{"citing_arxiv_id":"2606.12072","citing_title":"World Model Self-Distillation: Training World Models to Solve General Tasks","ref_index":33,"is_internal_anchor":true},{"citing_arxiv_id":"2606.03746","citing_title":"Qwen-Image-Flash: Beyond Objective Design","ref_index":4,"is_internal_anchor":true},{"citing_arxiv_id":"2606.02521","citing_title":"Drifting Preference Optimization for One-Step Generative Models","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2606.28144","citing_title":"Monocular Avatar Reconstruction via Cascaded Diffusion Priors and UV-Space Differentiable Shading","ref_index":27,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05204","citing_title":"D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14382","citing_title":"Delta Forcing: Trust Region Steering for Interactive Autoregressive Video Generation","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2605.25347","citing_title":"ERNIE-Image Technical Report","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2605.25378","citing_title":"CollectionLoRA: Collecting 50 Effects in 1 LoRA via Multi-Teacher On-Policy Distillation","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2605.26108","citing_title":"Reinforcing Few-step Generators via Reward-Tilted Distribution Matching","ref_index":29,"is_internal_anchor":true},{"citing_arxiv_id":"2605.21573","citing_title":"Lens: Rethinking Training Efficiency for Foundational Text-to-Image Models","ref_index":70,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14382","citing_title":"Delta Forcing: Trust Region Steering for Interactive Autoregressive Video Generation","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05204","citing_title":"D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14382","citing_title":"Delta Forcing: Trust Region Steering for Interactive Autoregressive Video Generation","ref_index":37,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15592","citing_title":"Efficient Image Synthesis with Sphere Latent Encoder","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14382","citing_title":"Delta Forcing: Trust Region Steering for Interactive Autoregressive Video Generation","ref_index":48,"is_internal_anchor":true},{"citing_arxiv_id":"2604.04018","citing_title":"1.x-Distill: Breaking the Diversity, Quality, and Efficiency Barrier in Distribution Matching Distillation","ref_index":14,"is_internal_anchor":true},{"citing_arxiv_id":"2605.03849","citing_title":"Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2604.23632","citing_title":"Hallo-Live: Real-Time Streaming Joint Audio-Video Avatar Generation with Asynchronous Dual-Stream and Human-Centric Preference Distillation","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.06376","citing_title":"Continuous-Time Distribution Matching for Few-Step Diffusion Distillation","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2605.05204","citing_title":"D-OPSD: On-Policy Self-Distillation for Continuously Tuning Step-Distilled Diffusion Models","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2511.22699","citing_title":"Z-Image: An Efficient Image Generation Foundation Model with Single-Stream Diffusion Transformer","ref_index":31,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO","json":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO.json","graph_json":"https://pith.science/api/pith-number/R6H35SOKFLENDXN3ADDGNKWUEO/graph.json","events_json":"https://pith.science/api/pith-number/R6H35SOKFLENDXN3ADDGNKWUEO/events.json","paper":"https://pith.science/paper/R6H35SOK"},"agent_actions":{"view_html":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO","download_json":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO.json","view_paper":"https://pith.science/paper/R6H35SOK","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2511.13649&json=true","fetch_graph":"https://pith.science/api/pith-number/R6H35SOKFLENDXN3ADDGNKWUEO/graph.json","fetch_events":"https://pith.science/api/pith-number/R6H35SOKFLENDXN3ADDGNKWUEO/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO/action/timestamp_anchor","attest_storage":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO/action/storage_attestation","attest_author":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO/action/author_attestation","sign_citation":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO/action/citation_signature","submit_replication":"https://pith.science/pith/R6H35SOKFLENDXN3ADDGNKWUEO/action/replication_record"}},"created_at":"2026-07-07T02:17:13.729907+00:00","updated_at":"2026-07-07T02:17:13.729907+00:00"}