Papers › eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers

eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoisers

2 Nov 2022arXiv:2211.01324archive 2025-07-28

Yogesh Balaji, Seungjun Nah, Xun Huang, Arash Vahdat, Jiaming Song, Qinsheng Zhang, Karsten Kreis, Miika Aittala, Timo Aila, Samuli Laine, Bryan Catanzaro, Tero Karras, Ming-Yu Liu

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 process may not be ideal. Therefore, in contrast to existing works, we propose to train an ensemble of text-to-image diffusion models specialized for different synthesis stages. To maintain training efficiency, we initially train a single model, which is then split into specialized models that are trained for the specific stages of the iterative generation process. 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. In addition, we train our model to exploit a variety of embeddings for conditioning, including the T5 text, CLIP text, and CLIP image embeddings. We show that these different embeddings lead to different behaviors. Notably, the CLIP image embedding allows an intuitive way of transferring the style of a reference image to the target text-to-image output. Lastly, we show a technique that enables eDiff-I's "paint-with-words" capability. A user can select the word in the input text and paint it in a canvas to control the output, which is very handy for crafting the desired image in mind. The project page is available at https://deepimagination.cc/eDiff-I/

PaperPDFCodeCode Syntology ran

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 13 code samples harvested from 2 repositories linked to this paper: 11 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 1 honoured, 0 violated, 9 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified. Where Syntology's instrument failed: 1 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: community (archive-listed): 13 samples from 2 repositories, 11 ran (10 with no instrument failure; 1 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

Levels written between 2026-09-22 and 2026-09-25 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

cloneofsimo/paint-with-words-sd mentioned on GitHubpytorchSyntology: 3 samples, 3 ran (1 instrument-failed) · 0 unverified report
cross-domain-compositing/cross-domain-compositing mentioned on GitHubpytorchMITSyntology: 10 samples, 8 ran (0 instrument-failed) · 2 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

13 samples harvested: 11 ran (of which 0 constructed an object rather than computing a result; 10 with no instrument failure: 1 honoured, 0 violated, 9 with no contract checked; 1 where Syntology's instrument failed) · 2 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
1ran; the contract check could not run
9ran
2unverified

Licence: 0 of the 13 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity. A census tag appears when Syntology's census ran implementations with that name on one shared input per bucket: it gives how many were compared, from how many papers, and how many distinct outputs they produced, and names this sample's bucket when this sample was one of them.

cloneofsimo/paint-with-words-sd community (archive-listed) 3 samples, 3 ran (1 instrument-failed) · 0 unverified
always_round paint_with_words/paint_with_words.py ran · honoured contract fingerprinted MIT (permissive) · a611fe4aef89ea52 · report fixture: signature x:float/0/float · returned int · level recorded 2026-09-22
inj_forward paint_with_words/paint_with_words.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant MIT (permissive) · de7114b637a2e1b6 · report fixture: signature self:callable/0/?, hidden_states:Tensor/3/float, context:dict/0/?, mask:NoneType/0/? · returned Tensor · level recorded 2026-09-22
preprocess in 284 papers
census: 11 implementations from 13 papers, 10 distinct outputs across 5 buckets, one shared input per bucket
paint_with_words/paint_with_words.py ran · metamorphic tier: well formed MIT (permissive) · cc073aabc2f49dab · report fixture: returned Tensor · level recorded 2026-09-22
cross-domain-compositing/cross-domain-compositing community (archive-listed) 10 samples, 8 ran (0 instrument-failed) · 2 unverified
cross_attention_masks cdc/prompt_masking.py ran MIT (permissive) · 99b6da8de85fe2bc · report fixture: signature mask:Tensor/4/float, tokens_in:list/1/int, tokens_out:list/1/int, w_in:float/0/float, w_out:float/0/float, max_tokens:int/0/int · returned Tensor · level recorded 2026-09-25
get_schedule in 6 papers cdc/scheduler.py ran MIT (permissive) · 314afb570b35e5c0 · report fixture: signature t_T:int/0/int, t_0:int/0/int, n_sample:int/0/int, n_steplength:int/0/int, debug:int/0/int · returned list · level recorded 2026-09-24
get_schedule_jump in 5 papers cdc/scheduler.py ran MIT (permissive) · aefaff0f568e5037 · report fixture: signature t_T:int/0/int, n_sample:int/0/int, jump_length:int/0/int, jump_n_sample:int/0/int, jump2_length:int/0/int, jump2_n_sample:int/0/int, jump3_length:int/0/int, jump3_n_sample:int/0/int, start_resampling:int/0/int, collapse_increasing:bool/0/bool · returned list · level recorded 2026-09-25
get_word_inds cdc/prompt_masking.py ran MIT (permissive) · 2032ab3924e8cd43 · report fixture: signature text:str/0/str, words:list/1/int, tokenizer:MockTokenizer/0/? · returned ndarray · level recorded 2026-09-25
load_checkpoint in 277 papers SVR/utils.py ran MIT (permissive) · 619771c3e74f395b · report fixture: signature cp_filename:str/0/str, model:callable/0/?, optimizer:SGD/0/? · returned tuple · level recorded 2026-09-25
load_model in 479 papers SVR/utils.py ran MIT (permissive) · f04426f214fbb1ec · report fixture: signature cp_filename:str/0/str, model:callable/0/? · returned tuple · level recorded 2026-09-25
process_image in 43 papers SVR/augment_ShapeNet_data.py ran MIT (permissive) · ce5593d971584146 · report fixture: signature image_file:str/0/str, save_path:str/0/str · returned tuple · level recorded 2026-09-25
sample_grid_points SVR/utils.py ran MIT (permissive) · 08767bfaed57712a · report fixture: signature xnum:int/0/int, ynum:int/0/int, znum:int/0/int · returned ndarray · level recorded 2026-09-25
download_models in 19 papers notebook_helpers.py unverified MIT (permissive) · 4533e43e06a12dc5 · report
get_blend_operator_and_mask cdc/cdc.py unverified MIT (permissive) · 0657ed0cbf2c4fc5 · report

Tasks

Image GenerationText-to-Image Generation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Text-to-Image Generation COCO (Common Objects in Context) eDiff-I (zero-shot) FID 6.95 #13 of 69 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AdafactorAttentionAttention DropoutBPECLIPDense ConnectionsDiffusionDropoutGated Linear UnitInverse Square Root ScheduleLayer NormalizationLinear LayerMulti-Head AttentionResidual ConnectionSentencePieceSoftmaxT5

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections