Papers › A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Data

A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Data

26 Feb 2024arXiv:2402.16991archive 2025-07-28

Antonio Sclocchi, Alessandro Favero, Matthieu Wyart

Understanding the structure of real data is paramount in advancing modern deep-learning methodologies. Natural data such as images are believed to be composed of features organized in a hierarchical and combinatorial manner, which neural networks capture during learning. Recent advancements show that diffusion models can generate high-quality images, hinting at their ability to capture this underlying compositional structure. We study this phenomenon in a hierarchical generative model of data. We find that the backward diffusion process acting after a time t is governed by a phase transition at some threshold time, where the probability of reconstructing high-level features, like the class of an image, suddenly drops. Instead, the reconstruction of low-level features, such as specific details of an image, evolves smoothly across the whole diffusion process. This result implies that at times beyond the transition, the class has changed, but the generated sample may still be composed of low-level elements of the initial image. We validate these theoretical insights through numerical experiments on class-unconditional ImageNet diffusion models. Our analysis characterizes the relationship between time and scale in diffusion models and puts forward generative models as powerful tools to model combinatorial data properties.

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 22 code samples harvested from 1 repository linked to this paper: 15 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 3 honoured, 0 violated, 0 with no contract checked; 12 where Syntology's instrument failed) · 7 unverified. Where Syntology's instrument failed: 10 where our contract draft was wrong; 2 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 22 samples from 1 repository, 15 ran (3 with no instrument failure; 12 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 2026-09-26 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed. 5 of the 15 samples that ran have no recorded date.

pcsl-epfl/forward-backward-diffusion officialmentioned in papermentioned on GitHubpytorchMITSyntology: 22 samples, 15 ran (12 instrument-failed) · 7 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

22 samples harvested: 15 ran (of which 0 constructed an object rather than computing a result; 3 with no instrument failure: 3 honoured, 0 violated, 0 with no contract checked; 12 where Syntology's instrument failed) · 7 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page last changed ; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

3ran · honoured contract
10ran · our draft was wrong
2ran; the contract check could not run
7unverified

Licence: 12 of the 22 samples are pointer only, meaning Syntology does not serve that copy's text. For 12 of them, this paper's copy is under a permissive licence and cleared, but the stored copy of the same code that Syntology serves is not; the licence label's tooltip gives Syntology's recorded reason. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from pcsl-epfl/forward-backward-diffusion. “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.

pcsl-epfl/forward-backward-diffusion official repository 22 samples, 15 ran (12 instrument-failed) · 7 unverified
center_crop_arr in 79 papers guided_diffusion/image_datasets.py ran · our draft was wrong MIT (permissive) · 8ab9a97df8161f31 · report fixture: signature pil_image:Image/0/?, image_size:int/0/int · returned ndarray · level recorded, date not recorded
approx_standard_normal_cdf in 150 papers
census: 4 implementations from 14 papers, 1 distinct output on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
guided_diffusion/losses.py ran · our draft was wrong fingerprinted MIT (permissive) here; the stored copy is pointer only · cfd76fd0d89574a4 · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded, date not recorded
betas_for_alpha_bar in 123 papers guided_diffusion/gaussian_diffusion.py ran · honoured contract MIT (permissive) here; the stored copy is pointer only · 2ab2316ac6fdd869 · report fixture: signature num_diffusion_timesteps:int/0/int, alpha_bar:callable/0/?, max_beta:float/0/float · returned ndarray · level recorded, date not recorded
create_named_schedule_sampler in 58 papers guided_diffusion/resample.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · 35572566acf61c78 · report fixture: signature name:str/0/str, diffusion:MockDiffusion/0/? · returned UniformSampler · level recorded 2026-09-22
discretized_gaussian_log_likelihood in 124 papers
census: 6 implementations from 10 papers, 2 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32 · arg 2: rank 2, kind float, dtype float32 · arg 3: rank 2, kind float, dtype float32
guided_diffusion/losses.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · cd33283d615fb3d7 · report fixture: signature x:Tensor/2/float, means:Tensor/2/float, log_scales:Tensor/2/float · returned Tensor · level recorded, date not recorded
get_named_beta_schedule in 119 papers guided_diffusion/gaussian_diffusion.py ran · honoured contract MIT (permissive) here; the stored copy is pointer only · a086d6286a40b889 · report fixture: signature schedule_name:str/0/str, num_diffusion_timesteps:int/0/int · returned ndarray · level recorded, date not recorded
get_param_groups_and_shapes in 73 papers guided_diffusion/fp16_util.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · e41367ad14ff58fd · report fixture: signature named_model_params:list/1/tuple · returned list · level recorded 2026-09-24
make_master_params in 90 papers guided_diffusion/fp16_util.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · e20dd5102da3b050 · report fixture: signature param_groups_and_shapes:list/1/tuple · returned list · level recorded 2026-09-24
make_output_format in 88 papers guided_diffusion/logger.py ran · our draft was wrong MIT (permissive) · bcd8b4acab199405 · report fixture: signature format:str/0/str, ev_dir:str/0/str, log_suffix:str/0/str · returned HumanOutputFormat · level recorded 2026-09-24
mpi_weighted_mean in 55 papers guided_diffusion/logger.py ran · our draft was wrong MIT (permissive) here; the stored copy is pointer only · e515a67f7f32e76d · report fixture: signature comm:MockComm/0/?, local_name2valcount:dict/0/? · returned dict · level recorded 2026-09-26
normal_kl in 188 papers
census: 14 implementations from 24 papers, 2 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32 · arg 2: rank 2, kind float, dtype float32 · arg 3: rank 2, kind float, dtype float32 · arg 4: rank 2, kind float, dtype float32
guided_diffusion/losses.py ran · honoured contract fingerprinted MIT (permissive) here; the stored copy is pointer only · cf2798b666b231ca · report fixture: signature mean1:Tensor/2/float, logvar1:Tensor/2/float, mean2:Tensor/2/float, logvar2:Tensor/2/float · returned Tensor · level recorded 2026-09-22
random_crop_arr in 41 papers guided_diffusion/image_datasets.py ran · our draft was wrong MIT (permissive) · 53d5be4fdbcfac23 · report fixture: signature pil_image:Image/0/?, image_size:int/0/int, min_crop_frac:float/0/float, max_crop_frac:float/0/float · returned ndarray · level recorded 2026-09-26
space_timesteps in 72 papers guided_diffusion/respace.py ran; the contract check could not run (RAISES) MIT (permissive) here; the stored copy is pointer only · 9b6e4e070473e76f · report fixture: signature num_timesteps:int/0/int, section_counts:list/1/int · returned set · level recorded 2026-09-22
unflatten_master_params in 90 papers guided_diffusion/fp16_util.py ran; the contract check could not run (RAISES) MIT (permissive) here; the stored copy is pointer only · 64fff1e30802b815 · report fixture: signature param_group:list/1/tuple, master_param:Tensor/1/float · returned tuple · level recorded 2026-09-24
zero_module in 184 papers guided_diffusion/nn.py ran · our draft was wrong MIT (permissive) · 129b804760b3115f · report fixture: signature module:callable/0/? · returned SimpleModule · level recorded 2026-09-22
avg_pool_nd in 83 papers guided_diffusion/nn.py unverified MIT (permissive) · ecd0fc28815b65ae · report
conv_nd in 94 papers guided_diffusion/nn.py unverified MIT (permissive) · fe4eb545bbb728e0 · report
create_classifier_and_diffusion in 11 papers guided_diffusion/script_util.py unverified MIT (permissive) · 5e5d99a81705a07f · report
create_model in 163 papers guided_diffusion/script_util.py unverified MIT (permissive) · 35ea054532112e53 · report
create_model_and_diffusion in 18 papers guided_diffusion/script_util.py unverified MIT (permissive) · 6a3cfc58dafbe392 · report
load_classifier in 7 papers guided_diffusion/torch_classifiers.py unverified MIT (permissive) · 2f10546095747e84 · report
profile in 89 papers guided_diffusion/logger.py unverified MIT (permissive) here; the stored copy is pointer only · 0c6607473a4c4c55 · report

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Diffusion

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