{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:3HE47GK55SEXOUKFRQT5D53FHA","short_pith_number":"pith:3HE47GK5","schema_version":"1.0","canonical_sha256":"d9c9cf995dec897751458c27d1f765381b103472802ad584455c6e5f211eea0d","source":{"kind":"arxiv","id":"2501.18427","version":4},"attestation_state":"computed","paper":{"title":"SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bingchen Liu, Chengyue Wu, Daquan Zhou, Enze Xie, Han Cai, Jincheng Yu, Junsong Chen, Junyu Chen, Ligeng Zhu, Muyang Li, Song Han, Yujun Lin, Yuyang Zhao, Zhekai Zhang","submitted_at":"2025-01-30T15:31:48Z","abstract_excerpt":"This paper presents SANA-1.5, a linear Diffusion Transformer for efficient scaling in text-to-image generation. Building upon SANA-1.0, we introduce three key innovations: (1) Efficient Training Scaling: A depth-growth paradigm that enables scaling from 1.6B to 4.8B parameters with significantly reduced computational resources, combined with a memory-efficient 8-bit optimizer. (2) Model Depth Pruning: A block importance analysis technique for efficient model compression to arbitrary sizes with minimal quality loss. (3) Inference-time Scaling: A repeated sampling strategy that trades computatio"},"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":"2501.18427","kind":"arxiv","version":4},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.CV","submitted_at":"2025-01-30T15:31:48Z","cross_cats_sorted":[],"title_canon_sha256":"845838c65c0172ce2eac4b748e23678622d659bb2caaf35f887c9c0c90f5c16c","abstract_canon_sha256":"791871af1d6b2cfbd7ef251f7ec17d0d33a7bed7e9efb3f5864a9fb436a449d2"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:04:44.685910Z","signature_b64":"MkWP8wGQjAl04U98aFySFFznQPI3DmMZTMbzeCE//k1LMK75DdVDqz3/ikZr6Icv4NYQDr3JY3+POGwaqddJBw==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"d9c9cf995dec897751458c27d1f765381b103472802ad584455c6e5f211eea0d","last_reissued_at":"2026-07-05T11:04:44.685363Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:04:44.685363Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"SANA 1.5: Efficient Scaling of Training-Time and Inference-Time Compute in Linear Diffusion Transformer","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Bingchen Liu, Chengyue Wu, Daquan Zhou, Enze Xie, Han Cai, Jincheng Yu, Junsong Chen, Junyu Chen, Ligeng Zhu, Muyang Li, Song Han, Yujun Lin, Yuyang Zhao, Zhekai Zhang","submitted_at":"2025-01-30T15:31:48Z","abstract_excerpt":"This paper presents SANA-1.5, a linear Diffusion Transformer for efficient scaling in text-to-image generation. Building upon SANA-1.0, we introduce three key innovations: (1) Efficient Training Scaling: A depth-growth paradigm that enables scaling from 1.6B to 4.8B parameters with significantly reduced computational resources, combined with a memory-efficient 8-bit optimizer. (2) Model Depth Pruning: A block importance analysis technique for efficient model compression to arbitrary sizes with minimal quality loss. (3) Inference-time Scaling: A repeated sampling strategy that trades computatio"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2501.18427","kind":"arxiv","version":4},"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/2501.18427/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":"2501.18427","created_at":"2026-07-05T11:04:44.685425+00:00"},{"alias_kind":"arxiv_version","alias_value":"2501.18427v4","created_at":"2026-07-05T11:04:44.685425+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2501.18427","created_at":"2026-07-05T11:04:44.685425+00:00"},{"alias_kind":"pith_short_12","alias_value":"3HE47GK55SEX","created_at":"2026-07-05T11:04:44.685425+00:00"},{"alias_kind":"pith_short_16","alias_value":"3HE47GK55SEXOUKF","created_at":"2026-07-05T11:04:44.685425+00:00"},{"alias_kind":"pith_short_8","alias_value":"3HE47GK5","created_at":"2026-07-05T11:04:44.685425+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":27,"internal_anchor_count":2,"sample":[{"citing_arxiv_id":"2607.06088","citing_title":"Flow Matching-Based Speech Source Separation with Best-of-N Biometric Sampling","ref_index":26,"is_internal_anchor":true},{"citing_arxiv_id":"2604.18518","citing_title":"UDM-GRPO: Stable and Efficient Group Relative Policy Optimization for Uniform Discrete Diffusion Models","ref_index":45,"is_internal_anchor":true},{"citing_arxiv_id":"2606.24849","citing_title":"IV-CoT: Implicit Visual Chain-of-Thought for Structure-Aware Text-to-Image Generation","ref_index":67,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27089","citing_title":"TMP: Tree-structured Mixed-policy Pruning for Large-scale Image Generation and Editing","ref_index":13,"is_internal_anchor":false},{"citing_arxiv_id":"2606.27371","citing_title":"Don't Settle at the Mode! Mitigating Diversity Collapse in Pretrained Flow Models via Feature Self-Guidance","ref_index":62,"is_internal_anchor":false},{"citing_arxiv_id":"2606.21030","citing_title":"FlowCodec: One-Step Flow Prior for Generative Image Compression","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19195","citing_title":"Moebius: 0.2B Lightweight Image Inpainting Framework with 10B-Level Performance","ref_index":52,"is_internal_anchor":false},{"citing_arxiv_id":"2606.13303","citing_title":"DuET: Dual Expert Trajectories for Diffusion Image Editing","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08231","citing_title":"Test-Time Scaling in Multimodal Foundation Models: A Comprehensive Survey of Generation and Reasoning","ref_index":96,"is_internal_anchor":false},{"citing_arxiv_id":"2606.06813","citing_title":"Breaking the Lock-in: Diversifying Text-to-Image Generation via Representation Modulation","ref_index":36,"is_internal_anchor":false},{"citing_arxiv_id":"2606.01803","citing_title":"OctoT2I: A Self-Evolving Agentic Text-to-Image Router","ref_index":54,"is_internal_anchor":false},{"citing_arxiv_id":"2606.30124","citing_title":"SciIR: A Large-scale Training Dataset and Benchmark for Scientific Image Reasoning Generation","ref_index":42,"is_internal_anchor":false},{"citing_arxiv_id":"2605.30409","citing_title":"SANA-Streaming: Real-time Streaming Video Editing with Hybrid Diffusion Transformer","ref_index":32,"is_internal_anchor":false},{"citing_arxiv_id":"2605.17500","citing_title":"The Silent Brush: Evaluating Artistic Style Leakage in AI Art Generation","ref_index":7,"is_internal_anchor":false},{"citing_arxiv_id":"2605.18115","citing_title":"WinTok: A Win-Win Hybrid Tokenizer via Decomposing Visual Understanding and Generation with Transferable Tokens","ref_index":94,"is_internal_anchor":false},{"citing_arxiv_id":"2605.19289","citing_title":"What Makes Synthetic Data Effective in Image Segmentation","ref_index":15,"is_internal_anchor":false},{"citing_arxiv_id":"2510.08431","citing_title":"Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency","ref_index":26,"is_internal_anchor":false},{"citing_arxiv_id":"2510.20206","citing_title":"RAPO++: Cross-Stage Prompt Optimization for Text-to-Video Generation via Data Alignment and Test-Time Scaling","ref_index":20,"is_internal_anchor":false},{"citing_arxiv_id":"2504.20690","citing_title":"In-Context Edit: Enabling Instructional Image Editing with In-Context Generation in Large Scale Diffusion Transformer","ref_index":60,"is_internal_anchor":false},{"citing_arxiv_id":"2502.12524","citing_title":"YOLOv12: Attention-Centric Real-Time Object Detectors","ref_index":63,"is_internal_anchor":false},{"citing_arxiv_id":"2605.11722","citing_title":"EPIC: Efficient Predicate-Guided Inference-Time Control for Compositional Text-to-Image Generation","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.12112","citing_title":"When Policy Entropy Constraint Fails: Preserving Diversity in Flow-based RLHF via Perceptual Entropy","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2505.05470","citing_title":"Flow-GRPO: Training Flow Matching Models via Online RL","ref_index":70,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12163","citing_title":"Nucleus-Image: Sparse MoE for Image Generation","ref_index":61,"is_internal_anchor":false},{"citing_arxiv_id":"2604.07340","citing_title":"TC-AE: Unlocking Token Capacity for Deep Compression Autoencoders","ref_index":11,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA","json":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA.json","graph_json":"https://pith.science/api/pith-number/3HE47GK55SEXOUKFRQT5D53FHA/graph.json","events_json":"https://pith.science/api/pith-number/3HE47GK55SEXOUKFRQT5D53FHA/events.json","paper":"https://pith.science/paper/3HE47GK5"},"agent_actions":{"view_html":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA","download_json":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA.json","view_paper":"https://pith.science/paper/3HE47GK5","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2501.18427&json=true","fetch_graph":"https://pith.science/api/pith-number/3HE47GK55SEXOUKFRQT5D53FHA/graph.json","fetch_events":"https://pith.science/api/pith-number/3HE47GK55SEXOUKFRQT5D53FHA/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA/action/timestamp_anchor","attest_storage":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA/action/storage_attestation","attest_author":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA/action/author_attestation","sign_citation":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA/action/citation_signature","submit_replication":"https://pith.science/pith/3HE47GK55SEXOUKFRQT5D53FHA/action/replication_record"}},"created_at":"2026-07-05T11:04:44.685425+00:00","updated_at":"2026-07-05T11:04:44.685425+00:00"}