{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2023:ROU4N4IRYZ3HV3M7HKTK3YY643","short_pith_number":"pith:ROU4N4IR","schema_version":"1.0","canonical_sha256":"8ba9c6f111c6767aed9f3aa6ade31ee6d7254c3ff5008fbdfaa60efc0bdb941f","source":{"kind":"arxiv","id":"2302.12192","version":1},"attestation_state":"computed","paper":{"title":"Aligning Text-to-Image Models using Human Feedback","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Fine-tuning text-to-image models with human feedback improves accuracy on prompts specifying colors, counts, and backgrounds.","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Craig Boutilier, Hao Liu, Kimin Lee, Mohammad Ghavamzadeh, Moonkyung Ryu, Olivia Watkins, Pieter Abbeel, Shixiang Shane Gu, Yuqing Du","submitted_at":"2023-02-23T17:34:53Z","abstract_excerpt":"Deep generative models have shown impressive results in text-to-image synthesis. However, current text-to-image models often generate images that are inadequately aligned with text prompts. We propose a fine-tuning method for aligning such models using human feedback, comprising three stages. First, we collect human feedback assessing model output alignment from a set of diverse text prompts. We then use the human-labeled image-text dataset to train a reward function that predicts human feedback. Lastly, the text-to-image model is fine-tuned by maximizing reward-weighted likelihood to improve "},"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":true,"formal_links_present":false},"canonical_record":{"source":{"id":"2302.12192","kind":"arxiv","version":1},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2023-02-23T17:34:53Z","cross_cats_sorted":["cs.AI","cs.CV"],"title_canon_sha256":"b9f9d0a75704678b4acde8885bfbea84f3cb70eca4a6315ff07a9f7d2fb3b1f0","abstract_canon_sha256":"2c26c5b752bc8facff7879651ec9d50c096133e39800c4894b64e41096fcbb67"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-05-17T23:39:21.805985Z","signature_b64":"hEJlGv0W8AGsaypS5YJB1qcRGiTXbG8+8AENeAzCX4GQwsYfW6WSQPKRItl24i6O2yDFarRK/1nIvE6aocIBDw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8ba9c6f111c6767aed9f3aa6ade31ee6d7254c3ff5008fbdfaa60efc0bdb941f","last_reissued_at":"2026-05-17T23:39:21.805215Z","signature_status":"signed_v1","first_computed_at":"2026-05-17T23:39:21.805215Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Aligning Text-to-Image Models using Human Feedback","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"Fine-tuning text-to-image models with human feedback improves accuracy on prompts specifying colors, counts, and backgrounds.","cross_cats":["cs.AI","cs.CV"],"primary_cat":"cs.LG","authors_text":"Craig Boutilier, Hao Liu, Kimin Lee, Mohammad Ghavamzadeh, Moonkyung Ryu, Olivia Watkins, Pieter Abbeel, Shixiang Shane Gu, Yuqing Du","submitted_at":"2023-02-23T17:34:53Z","abstract_excerpt":"Deep generative models have shown impressive results in text-to-image synthesis. However, current text-to-image models often generate images that are inadequately aligned with text prompts. We propose a fine-tuning method for aligning such models using human feedback, comprising three stages. First, we collect human feedback assessing model output alignment from a set of diverse text prompts. We then use the human-labeled image-text dataset to train a reward function that predicts human feedback. Lastly, the text-to-image model is fine-tuned by maximizing reward-weighted likelihood to improve "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"Our method generates objects with specified colors, counts and backgrounds more accurately than the pre-trained model.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"Human feedback on image-text alignment is consistent enough to be captured by a learned reward function that generalizes to new prompts and does not introduce unintended biases during fine-tuning.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A three-stage fine-tuning process uses human ratings to train a reward model and then improves text-to-image alignment by maximizing reward-weighted likelihood.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Fine-tuning text-to-image models with human feedback improves accuracy on prompts specifying colors, counts, and backgrounds.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"6cd0b1bf33b50df1bbf267ede71a4204cbb4f4fe5e7600cee531f564453ec6e1"},"source":{"id":"2302.12192","kind":"arxiv","version":1},"verdict":{"id":"88c2a63e-42c0-43e5-86ce-ba86a719a11a","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T20:35:53.230683Z","strongest_claim":"Our method generates objects with specified colors, counts and backgrounds more accurately than the pre-trained model.","one_line_summary":"A three-stage fine-tuning process uses human ratings to train a reward model and then improves text-to-image alignment by maximizing reward-weighted likelihood.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"Human feedback on image-text alignment is consistent enough to be captured by a learned reward function that generalizes to new prompts and does not introduce unintended biases during fine-tuning.","pith_extraction_headline":"Fine-tuning text-to-image models with human feedback improves accuracy on prompts specifying colors, counts, and backgrounds."},"references":{"count":26,"sample":[{"doi":"","year":null,"title":"A General Language Assistant as a Laboratory for Alignment","work_id":"a43f9ea0-01be-47d5-b8ee-a1a9f73381c5","ref_index":1,"cited_arxiv_id":"2112.00861","is_internal_anchor":true},{"doi":"","year":null,"title":"arXiv preprint arXiv:1607.07086 , year=","work_id":"a298def3-ff4b-4b72-9b55-5707acf335ba","ref_index":2,"cited_arxiv_id":"1607.07086","is_internal_anchor":true},{"doi":"","year":2005,"title":"Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback","work_id":"a1f2574b-a899-4713-be60-c87ba332656c","ref_index":3,"cited_arxiv_id":"2204.05862","is_internal_anchor":true},{"doi":"","year":null,"title":"E., and Wang, W","work_id":"63b29782-c84e-4467-9583-aee36944de57","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":null,"title":"An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversion","work_id":"ca618c21-3ba6-448e-bd86-bcecff3cdeb5","ref_index":5,"cited_arxiv_id":"2208.01618","is_internal_anchor":true}],"resolved_work":26,"snapshot_sha256":"86521c9baf0ae9ec1490431a238d8f1e0c731923e8d77dff6e41c6f4033768ab","internal_anchors":17},"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":"2302.12192","created_at":"2026-05-17T23:39:21.805345+00:00"},{"alias_kind":"arxiv_version","alias_value":"2302.12192v1","created_at":"2026-05-17T23:39:21.805345+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2302.12192","created_at":"2026-05-17T23:39:21.805345+00:00"},{"alias_kind":"pith_short_12","alias_value":"ROU4N4IRYZ3H","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_16","alias_value":"ROU4N4IRYZ3HV3M7","created_at":"2026-05-18T12:33:37.589309+00:00"},{"alias_kind":"pith_short_8","alias_value":"ROU4N4IR","created_at":"2026-05-18T12:33:37.589309+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":39,"internal_anchor_count":39,"sample":[{"citing_arxiv_id":"2503.12575","citing_title":"BalancedDPO: Adaptive Multi-Metric Alignment","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2505.17352","citing_title":"Alignment and Safety of Diffusion Models via Reinforcement Learning and Reward Modeling: A Survey","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2602.22801","citing_title":"Unleashing the Potential of Diffusion Models for End-to-End Autonomous Driving","ref_index":25,"is_internal_anchor":true},{"citing_arxiv_id":"2605.02439","citing_title":"Anomaly-Preference Image Generation","ref_index":23,"is_internal_anchor":true},{"citing_arxiv_id":"2605.15803","citing_title":"Embedding-perturbed Exploration Preference Optimization for Flow Models","ref_index":42,"is_internal_anchor":true},{"citing_arxiv_id":"2406.03520","citing_title":"VideoPhy: Evaluating Physical Commonsense for Video Generation","ref_index":52,"is_internal_anchor":true},{"citing_arxiv_id":"2605.19750","citing_title":"CPC-VAR:Continual Personalized and Compositional Generation in Visual Autoregressive Models","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2605.16937","citing_title":"DEVIS-GRPO: Unleashing GRPO on Dynamic Extreme View Synthesis","ref_index":72,"is_internal_anchor":true},{"citing_arxiv_id":"2509.25844","citing_title":"Believing without Seeing: Quality Scores for Contextualizing Vision-Language Model Explanations","ref_index":2,"is_internal_anchor":true},{"citing_arxiv_id":"2510.20093","citing_title":"StableSketcher: Enhancing Diffusion Model for Pixel-based Sketch Generation via Visual Question Answering Feedback","ref_index":16,"is_internal_anchor":true},{"citing_arxiv_id":"2511.00088","citing_title":"Alpamayo-R1: Bridging Reasoning and Action Prediction for Generalizable Autonomous Driving in the Long Tail","ref_index":42,"is_internal_anchor":true},{"citing_arxiv_id":"2304.06767","citing_title":"RAFT: Reward rAnked FineTuning for Generative Foundation Model Alignment","ref_index":84,"is_internal_anchor":true},{"citing_arxiv_id":"2512.04678","citing_title":"Reward Forcing: Efficient Streaming Video Generation with Rewarded Distribution Matching Distillation","ref_index":38,"is_internal_anchor":true},{"citing_arxiv_id":"2309.17400","citing_title":"Directly Fine-Tuning Diffusion Models on Differentiable Rewards","ref_index":12,"is_internal_anchor":true},{"citing_arxiv_id":"2602.04713","citing_title":"Adaptive Prompt Elicitation for Text-to-Image Generation","ref_index":49,"is_internal_anchor":true},{"citing_arxiv_id":"2603.00918","citing_title":"Improving Text-to-Image Generation with Intrinsic Self-Confidence Rewards","ref_index":32,"is_internal_anchor":true},{"citing_arxiv_id":"2603.07119","citing_title":"TIQA: Human-Aligned Perceptual Text Quality Assessment in Generated Images","ref_index":34,"is_internal_anchor":true},{"citing_arxiv_id":"2605.14274","citing_title":"CreFlow: Corrective Reflow for Sparse-Reward Embodied Video Diffusion RL","ref_index":19,"is_internal_anchor":true},{"citing_arxiv_id":"2503.05236","citing_title":"Unified Reward Model for Multimodal Understanding and Generation","ref_index":8,"is_internal_anchor":true},{"citing_arxiv_id":"2509.16117","citing_title":"DiffusionNFT: Online Diffusion Reinforcement with Forward Process","ref_index":10,"is_internal_anchor":true},{"citing_arxiv_id":"2501.13918","citing_title":"Improving Video Generation with Human Feedback","ref_index":36,"is_internal_anchor":true},{"citing_arxiv_id":"2507.21802","citing_title":"MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE","ref_index":15,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09125","citing_title":"Transfer Learning of Multiobjective Indirect Low-Thrust Trajectories Using Diffusion Models and Markov Chain Monte Carlo","ref_index":39,"is_internal_anchor":true},{"citing_arxiv_id":"2605.10937","citing_title":"Power Reinforcement Post-Training of Text-to-Image Models with Super-Linear Advantage Shaping","ref_index":77,"is_internal_anchor":true},{"citing_arxiv_id":"2605.09291","citing_title":"dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models","ref_index":130,"is_internal_anchor":true}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643","json":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643.json","graph_json":"https://pith.science/api/pith-number/ROU4N4IRYZ3HV3M7HKTK3YY643/graph.json","events_json":"https://pith.science/api/pith-number/ROU4N4IRYZ3HV3M7HKTK3YY643/events.json","paper":"https://pith.science/paper/ROU4N4IR"},"agent_actions":{"view_html":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643","download_json":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643.json","view_paper":"https://pith.science/paper/ROU4N4IR","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2302.12192&json=true","fetch_graph":"https://pith.science/api/pith-number/ROU4N4IRYZ3HV3M7HKTK3YY643/graph.json","fetch_events":"https://pith.science/api/pith-number/ROU4N4IRYZ3HV3M7HKTK3YY643/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643/action/timestamp_anchor","attest_storage":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643/action/storage_attestation","attest_author":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643/action/author_attestation","sign_citation":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643/action/citation_signature","submit_replication":"https://pith.science/pith/ROU4N4IRYZ3HV3M7HKTK3YY643/action/replication_record"}},"created_at":"2026-05-17T23:39:21.805345+00:00","updated_at":"2026-05-17T23:39:21.805345+00:00"}