{"state_type":"pith_open_graph_state","state_version":"1.0","pith_number":"pith:2022:Y5HR3MRXMK6FSX6PMAM4S5Z6Q4","merge_version":"pith-open-graph-merge-v1","event_count":2,"valid_event_count":2,"invalid_event_count":0,"equivocation_count":0,"current":{"canonical_record":{"metadata":{"abstract_canon_sha256":"6ba932880558e507ebb983d38e510f4e566f11bba98dcd82cf7a99666372929f","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2022-11-02T17:43:04Z","title_canon_sha256":"e9f8d1e62e63e995eafd74d40a92cab36ae1e21fca1d6f3683a0e8e5ac675eea"},"schema_version":"1.0","source":{"id":"2211.01324","kind":"arxiv","version":5}},"source_aliases":[{"alias_kind":"arxiv","alias_value":"2211.01324","created_at":"2026-05-17T23:39:05Z"},{"alias_kind":"arxiv_version","alias_value":"2211.01324v5","created_at":"2026-05-17T23:39:05Z"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2211.01324","created_at":"2026-05-17T23:39:05Z"},{"alias_kind":"pith_short_12","alias_value":"Y5HR3MRXMK6F","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_16","alias_value":"Y5HR3MRXMK6FSX6P","created_at":"2026-05-18T12:33:33Z"},{"alias_kind":"pith_short_8","alias_value":"Y5HR3MRX","created_at":"2026-05-18T12:33:33Z"}],"graph_snapshots":[{"event_id":"sha256:f743f77829b3bf973f5bc9117733dd3bd4d449b74507b29511eac1fd9d7016f3","target":"graph","created_at":"2026-05-17T23:39:05Z","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":4,"items":[{"attestation":"unclaimed","claim_id":"C1","kind":"strongest_claim","source":"verdict.strongest_claim","status":"machine_extracted","text":"Our ensemble of diffusion models, called eDiff-I, results in improved text alignment while maintaining the same inference computation cost and preserving high visual quality, outperforming previous large-scale text-to-image diffusion models on the standard benchmark."},{"attestation":"unclaimed","claim_id":"C2","kind":"weakest_assumption","source":"verdict.weakest_assumption","status":"machine_extracted","text":"The synthesis behavior qualitatively changes throughout the generation process such that early stages rely on text conditioning while later stages largely ignore it, making a single shared-parameter model suboptimal."},{"attestation":"unclaimed","claim_id":"C3","kind":"one_line_summary","source":"verdict.one_line_summary","status":"machine_extracted","text":"An ensemble of stage-specialized text-to-image diffusion models improves prompt alignment over single shared-parameter models while preserving visual quality and inference speed."},{"attestation":"unclaimed","claim_id":"C4","kind":"headline","source":"verdict.pith_extraction.headline","status":"machine_extracted","text":"An ensemble of stage-specialized diffusion models improves text alignment in image synthesis at the same inference cost."}],"snapshot_sha256":"d19838852ca4de31ef0d83cb2cbcd3555be749baa7f1b69390136d0a2a8993bb"},"formal_canon":{"evidence_count":2,"snapshot_sha256":"25a2d4e30e413f2993b5b0dc4bbf4653a831193a26daf8e758a4a4c926d03d5a"},"paper":{"abstract_excerpt":"Large-scale diffusion-based generative models have led to breakthroughs in text-conditioned high-resolution image synthesis. Starting from random noise, such text-to-image diffusion models gradually synthesize images in an iterative fashion while conditioning on text prompts. We find that their synthesis behavior qualitatively changes throughout this process: Early in sampling, generation strongly relies on the text prompt to generate text-aligned content, while later, the text conditioning is almost entirely ignored. This suggests that sharing model parameters throughout the entire generation","authors_text":"Arash Vahdat, Bryan Catanzaro, Jiaming Song, Karsten Kreis, Miika Aittala, Ming-Yu Liu, Qinsheng Zhang, Samuli Laine, Seungjun Nah, Tero Karras, Timo Aila, Xun Huang, Yogesh Balaji","cross_cats":["cs.LG"],"headline":"An ensemble of stage-specialized diffusion models improves text alignment in image synthesis at the same inference cost.","license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2022-11-02T17:43:04Z","title":"eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers"},"references":{"count":95,"internal_anchors":11,"resolved_work":95,"sample":[{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":1,"title":"V ., Du, J., Iyer, S., Pasunuru, R., et al","work_id":"711d98e0-b9e1-4350-b81d-a621de85a6bb","year":2021},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":2,"title":"Blended latent diffusion","work_id":"1aef642e-c5e8-42d8-a391-f41998245355","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":3,"title":"Blended diffusion for text-driven editing of natural images","work_id":"b7d69bd8-2483-45c0-93f6-0bf59101b935","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":4,"title":"Estimating the optimal covariance with imperfect mean in diffusion probabilistic models","work_id":"6a02cf87-87e3-4e01-8807-1815408583fe","year":2022},{"cited_arxiv_id":"","doi":"","is_internal_anchor":false,"ref_index":5,"title":"Analytic- DPM: An analytic estimate of the optimal reverse variance in diffusion probabilistic models","work_id":"83033437-6996-4dfd-9f0f-41f417388b99","year":2022}],"snapshot_sha256":"534b6679319b65d105d3f306b7c44eee8a40b75cbb5a95d217ab7b4a06494e91"},"source":{"id":"2211.01324","kind":"arxiv","version":5},"verdict":{"created_at":"2026-05-15T01:40:27.202143Z","id":"ea0e3657-7f80-4a78-848a-196756727a3f","model_set":{"reader":"grok-4.3"},"one_line_summary":"An ensemble of stage-specialized text-to-image diffusion models improves prompt alignment over single shared-parameter models while preserving visual quality and inference speed.","pipeline_version":"pith-pipeline@v0.9.0","pith_extraction_headline":"An ensemble of stage-specialized diffusion models improves text alignment in image synthesis at the same inference cost.","strongest_claim":"Our ensemble of diffusion models, called eDiff-I, results in improved text alignment while maintaining the same inference computation cost and preserving high visual quality, outperforming previous large-scale text-to-image diffusion models on the standard benchmark.","weakest_assumption":"The synthesis behavior qualitatively changes throughout the generation process such that early stages rely on text conditioning while later stages largely ignore it, making a single shared-parameter model suboptimal."}},"verdict_id":"ea0e3657-7f80-4a78-848a-196756727a3f"}}],"author_attestations":[],"timestamp_anchors":[],"storage_attestations":[],"citation_signatures":[],"replication_records":[],"corrections":[],"mirror_hints":[],"record_created":{"event_id":"sha256:5ee1e5c5fb7dcb0c09f436f604809d2a8f2ebb9c81102e7a002d8ed92c3a7b7b","target":"record","created_at":"2026-05-17T23:39:05Z","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":"6ba932880558e507ebb983d38e510f4e566f11bba98dcd82cf7a99666372929f","cross_cats_sorted":["cs.LG"],"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2022-11-02T17:43:04Z","title_canon_sha256":"e9f8d1e62e63e995eafd74d40a92cab36ae1e21fca1d6f3683a0e8e5ac675eea"},"schema_version":"1.0","source":{"id":"2211.01324","kind":"arxiv","version":5}},"canonical_sha256":"c74f1db23762bc595fcf6019c9773e8727f464891dc08adab64526d371aebe71","receipt":{"algorithm":"ed25519","builder_version":"pith-number-builder-2026-05-17-v1","canonical_sha256":"c74f1db23762bc595fcf6019c9773e8727f464891dc08adab64526d371aebe71","first_computed_at":"2026-05-17T23:39:05.095621Z","key_id":"pith-v1-2026-05","kind":"pith_receipt","last_reissued_at":"2026-05-17T23:39:05.095621Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54","receipt_version":"0.3","signature_b64":"IjBzyxBOwfyFIG43mI/vCHaKLPtENn6mPd6ux2lYUUJ0K5R57wF5bF9IR3+aJ93AU7KqfBJJOe/H95rkfmtHBw==","signature_status":"signed_v1","signed_at":"2026-05-17T23:39:05.096476Z","signed_message":"canonical_sha256_bytes"},"source_id":"2211.01324","source_kind":"arxiv","source_version":5}}},"equivocations":[],"invalid_events":[],"applied_event_ids":["sha256:5ee1e5c5fb7dcb0c09f436f604809d2a8f2ebb9c81102e7a002d8ed92c3a7b7b","sha256:f743f77829b3bf973f5bc9117733dd3bd4d449b74507b29511eac1fd9d7016f3"],"state_sha256":"ad263188ea371b644de42fb609e248e2b5deed69fad2b5bae4a9c053439a8c36"}