Papers › Argmax Flows and Multinomial Diffusion: Learning Categorical Distributions

Argmax Flows and Multinomial Diffusion: Learning Categorical Distributions

10 Feb 2021NeurIPS 2021 12arXiv:2102.05379archive 2025-07-28

Emiel Hoogeboom, Didrik Nielsen, Priyank Jaini, Patrick Forré, Max Welling

Generative flows and diffusion models have been predominantly trained on ordinal data, for example natural images. This paper introduces two extensions of flows and diffusion for categorical data such as language or image segmentation: Argmax Flows and Multinomial Diffusion. Argmax Flows are defined by a composition of a continuous distribution (such as a normalizing flow), and an argmax function. To optimize this model, we learn a probabilistic inverse for the argmax that lifts the categorical data to a continuous space. Multinomial Diffusion gradually adds categorical noise in a diffusion process, for which the generative denoising process is learned. We demonstrate that our method outperforms existing dequantization approaches on text modelling and modelling on image segmentation maps in log-likelihood.

PaperPDFConference PDFCodeCode Syntology ranMethod as code

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.

Syntology holds a reference implementation of Argmax Flows, generated from the paper text (not the authors' code), at level 3 on Syntology's ladder: V2 “two implementations GENERATED from the paper's signature and spec agree numerically (1e-6) on fixture inputs” “neither of the two ever read the authors' own code, so V2 measures how unambiguously the paper specifies the method, not fidelity to the authors' implementation (measured 2026-09-11, commit 75eaa77: cross_checked_by is a second LLM on 2,628 of 2,831 served samples)”; V3 “paper-derived property tests pass on both of those same two generations.” (quoted from Syntology's server card). For this implementation, level 3 is recorded, but its report lists no property test that passed on both. No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='Argmax Flows') (how to connect); the free check have('2102.05379') needs no token.

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

Code

Syntology 20 code samples harvested from 4 repositories linked to this paper: 14 ran (of which 1 constructed an object rather than computing a result; 9 with no instrument failure: 7 honoured, 1 violated, 1 with no contract checked; 5 where Syntology's instrument failed) · 6 unverified. Where Syntology's instrument failed: 2 where our contract draft was wrong; 3 where the contract check could not run; failures of Syntology's instrument, not of the code.

By repository: official repository: 12 samples from 2 repositories, 9 ran (6 with no instrument failure; 3 where Syntology's instrument failed); community (archive-listed): 7 samples from 1 repository, 5 ran (3 with no instrument failure; 2 where Syntology's instrument failed); found in paper text by Syntology: 1 sample from 1 repository, 0 ran (0 with no instrument failure; 0 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 on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

didriknielsen/argmax_flows officialmentioned on GitHubpytorchSyntology: 11 samples, 8 ran (3 instrument-failed) · 3 unverified report
ehoogeboom/multinomial_diffusion officialmentioned on GitHubpytorchSyntology: 1 sample, 1 ran (0 instrument-failed) · 0 unverified report
YuzheZhang-1999/DiffTSR mentioned on GitHubpytorchSyntology: 7 samples, 5 ran (2 instrument-failed) · 2 unverified report
compvis/imagebart mentioned on GitHubpytorchSyntology: not harvested 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

20 samples harvested: 14 ran (of which 1 constructed an object rather than computing a result; 9 with no instrument failure: 7 honoured, 1 violated, 1 with no contract checked; 5 where Syntology's instrument failed) · 6 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.

7ran · honoured contract
1ran · violated contract
2ran · our draft was wrong
3ran; the contract check could not run
1ran
6unverified

Licence: 2 of the 20 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 4 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.

didriknielsen/argmax_flows official repository 11 samples, 8 ran (3 instrument-failed) · 3 unverified
_log_cdf model/CategoricalNF/flows/mixture_cdf_layer.py ran · honoured contract fingerprinted MIT (permissive) · e72047c1892c7a2a · report fixture: signature x:Tensor/2/float, mean:Tensor/2/float, log_scale:Tensor/2/float · returned Tensor · level recorded 2026-09-22
_log_pdf model/CategoricalNF/flows/mixture_cdf_layer.py ran · honoured contract fingerprinted MIT (permissive) · 74b5245783db5aab · report fixture: signature x:Tensor/2/float, mean:Tensor/2/float, log_scale:Tensor/2/float · returned Tensor · level recorded 2026-09-22
inverse in 15 papers model/CategoricalNF/flows/mixture_cdf_layer.py ran; the contract check could not run (RAISES) fingerprinted MIT (permissive) · 7e630357d1081630 · report fixture: signature x:Tensor/2/float, reverse:bool/0/bool · returned tuple · level recorded 2026-09-22
mixture_inv_cdf model/CategoricalNF/flows/mixture_cdf_layer.py ran; the contract check could not run (RAISES) · metamorphic tier: invariant MIT (permissive) · cf94102882d34c5a · report fixture: signature y:Tensor/1/float, prior_logits:Tensor/2/float, means:Tensor/2/float, log_scales:Tensor/2/float, eps:float/0/float, max_iters:int/0/int · returned Tensor · level recorded 2026-09-22
mixture_log_cdf model/CategoricalNF/flows/mixture_cdf_layer.py ran · violated contract fingerprinted MIT (permissive) · db244b4a1b9d7755 · report fixture: signature x:Tensor/1/float, prior_logits:Tensor/1/float, means:Tensor/1/float, log_scales:Tensor/1/float · returned Tensor · level recorded 2026-09-22
mixture_log_pdf model/CategoricalNF/flows/mixture_cdf_layer.py ran · honoured contract fingerprinted MIT (permissive) · 3ae2ea791a983a23 · report fixture: signature x:Tensor/2/float, prior_logits:Tensor/1/float, means:Tensor/1/float, log_scales:Tensor/1/float · returned Tensor · level recorded 2026-09-22
run_sequential_with_mask model/CategoricalNF/flows/mixture_cdf_layer.py ran · our draft was wrong MIT (permissive) · 3ee1217b23068736 · report fixture: signature net:callable/0/?, x:Tensor/3/float, length:Tensor/1/int, channel_padding_mask:Tensor/3/float, src_key_padding_mask:Tensor/2/bool, length_one_hot:Tensor/2/float, time_embed:Tensor/3/float, gt:Tensor/3/float, importance_weight:float/0/float, detail_out:bool/0/bool · returned Tensor · level recorded 2026-09-22
safe_log in 27 papers
census: 5 implementations from 5 papers, 3 distinct outputs on one shared input this sample: bucket 1 of 1, arg 1: rank 2, kind float, dtype float32
model/CategoricalNF/flows/mixture_cdf_layer.py ran · honoured contract fingerprinted MIT (permissive) · 13bc23344ef8bd53 · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded 2026-09-22
CouplingLayer in 11 papers model/CategoricalNF/flows/mixture_cdf_layer.py unverified MIT (permissive) · 62906ae69863334e · report
FlowLayer model/CategoricalNF/flows/mixture_cdf_layer.py unverified MIT (permissive) · 0fd897703533ac05 · report
MixtureCDFCoupling model/CategoricalNF/flows/mixture_cdf_layer.py unverified MIT (permissive) · 790a25da85a326bc · report
ehoogeboom/multinomial_diffusion official repository 1 sample, 1 ran (0 instrument-failed) · 0 unverified
MultinomialDiffusion diffusion_utils/diffusion_multinomial.py ran · metamorphic tier: deterministic no licence file found · pointer only · 4268dcf6913cd21f · report fixture: returned MultinomialDiffusion · level recorded 2026-09-22
YuzheZhang-1999/DiffTSR community (archive-listed) 7 samples, 5 ran (2 instrument-failed) · 2 unverified
cosine_beta_schedule in 76 papers model/TDM/models/diffusion_multinomial.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · c92a02b48d8fb1c4 · report fixture: signature timesteps:int/0/int, s:float/0/float · returned ndarray · level recorded 2026-09-22
index_to_log_onehot in 7 papers model/TDM/models/diffusion_multinomial.py ran; the contract check could not run (RAISES) Apache-2.0 (permissive) · 887b510303aa0cff · report fixture: signature x:Tensor/3/int, num_classes:int/0/int · returned Tensor · level recorded 2026-09-22
log_1_min_a in 13 papers model/TDM/models/diffusion_multinomial.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · 39226fe0563dc032 · report fixture: signature a:Tensor/2/float · returned Tensor · level recorded 2026-09-22
log_add_exp in 12 papers model/TDM/models/diffusion_multinomial.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · 3d24f690340703af · report fixture: signature a:Tensor/2/float, b:Tensor/2/float · returned Tensor · level recorded 2026-09-22
sum_except_batch in 25 papers
census: 3 implementations from 5 papers, 3 distinct outputs across 3 buckets, one shared input per bucket this sample: bucket 2 of 3, arg 1: rank 3, kind float, dtype float32
model/TDM/models/diffusion_multinomial.py ran · our draft was wrong Apache-2.0 (permissive) · 38d8ef27c0f4b98a · report fixture: signature x:Tensor/3/float, num_dims:int/0/int · returned Tensor · level recorded 2026-09-22
DiffusionWrapper in 5 papers model/TDM/models/diffusion_multinomial.py unverified Apache-2.0 (permissive) · a15ee5030a6486ce · report
MultinomialDiffusion model/TDM/models/diffusion_multinomial.py unverified Apache-2.0 (permissive) · 08aec1a58f72176a · report
TrentBrick/PyTorchDiscreteFlows found in paper text by Syntology 1 sample, 0 ran · 1 unverified
DiscreteAutoregressiveFlow discrete_flows/disc_models.py unverified no licence file found · pointer only · b2a81908ae5a22a5 · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. For a generated assembly, an independent peer is a second implementation generated separately from the same paper text; neither implementation read the authors' code. At level 2 the two agreed numerically on the fixture's inputs; level 3 is recorded when property tests derived from the paper also passed on both, and each level 3 card prints how many its report lists. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. A generated assembly's label was set when its code was generated for the method, and no model judged it; its card prints it as "by construction" or "not judged by a model". Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another. Where an assembly's phrase has no constructed figure, a brick that ran is not among the harvested samples this build reads, and whether a brick's run only constructed an object is read from that brick's harvested sample, so the figure was not counted; it does not say there were none.

5 assemblies on this paper: 4 harvested, 1 generated, listed in 3 groups (siblings lifted in one batch form one group).

generated Assembly for Argmax Flows (graph method: Argmax Flows), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (1 with no instrument failure: 1 honoured, 0 violated, 0 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer; level 3 is recorded, but its report lists no property test that passed on both.

label matches by construction: the code was generated for this method (not judged by a model)

  1. compute_argmax_flow_elbo entry · level 3 · ran · honoured contract · a3ca60847973f6c4035839ea7d14231c8140a723ce09af51669c0a85232fbae0

2 sibling assemblies for Argmax Flows (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for Argmax Flows (graph method: Argmax Flows). 11 bricks; individually on a fixture: 8 ran (of which 0 constructed an object rather than computing a result; 5 with no instrument failure: 4 honoured, 1 violated, 0 with no contract checked; 3 where Syntology's instrument failed) · 3 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. MixtureCDFCoupling entry · level 0 · unverified · 790a25da85a326bc
  2. CouplingLayer helper · level 0 · unverified · 62906ae69863334e
  3. inverse model · level 1 · ran; the contract check could not run (RAISES) · 7e630357d1081630
  4. mixture_inv_cdf model · level 1 · ran; the contract check could not run (RAISES) · cf94102882d34c5a
  5. mixture_log_cdf model · level 1 · ran · violated contract · db244b4a1b9d7755
  6. mixture_log_pdf model · level 1 · ran · honoured contract · 3ae2ea791a983a23
  7. safe_log helper · level 1 · ran · honoured contract · 13bc23344ef8bd53
  8. FlowLayer helper · level 0 · unverified · 0fd897703533ac05
  9. run_sequential_with_mask helper · level 1 · ran · our draft was wrong · 3ee1217b23068736
  10. _log_cdf helper · level 1 · ran · honoured contract · e72047c1892c7a2a
  11. _log_pdf helper · level 1 · ran · honoured contract · 74b5245783db5aab

harvested Assembly for Argmax Flows (graph method: Argmax Flows). 1 brick; individually on a fixture: 0 ran · 1 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. DiscreteAutoregressiveFlow entry · level 0 · unverified · b2a81908ae5a22a5

2 sibling assemblies for Multinomial Diffusion (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for Multinomial Diffusion (graph method: Multinomial Diffusion). 13 bricks; individually on a fixture: 9 ran (of which 0 constructed an object rather than computing a result; 6 with no instrument failure: 5 honoured, 1 violated, 0 with no contract checked; 3 where Syntology's instrument failed) · 4 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. MultinomialDiffusion entry · level 0 · unverified · 08aec1a58f72176a
  2. DiffusionWrapper helper · level 0 · unverified · a15ee5030a6486ce
  3. cosine_beta_schedule scheduler · level 1 · ran · honoured contract · c92a02b48d8fb1c4
  4. disabled_train helper · level 1 · ran · violated contract · 4cb732f513d69dfd
  5. extract helper · level 0 · unverified · 09c8479d9a5b3e06
  6. index_to_log_onehot data_transform · level 1 · ran; the contract check could not run (RAISES) · 887b510303aa0cff
  7. instantiate_from_config helper · level 0 · unverified · ef52d142b2c30ac2
  8. log_1_min_a helper · level 1 · ran · honoured contract · 39226fe0563dc032
  9. log_add_exp helper · level 1 · ran · honoured contract · 3d24f690340703af
  10. log_categorical loss · level 1 · ran · honoured contract · f4b18dd8f67674cb
  11. log_onehot_to_index data_transform · level 1 · ran · honoured contract · 5def5f79a41cfc04
  12. sum_except_batch helper · level 1 · ran · our draft was wrong · 38d8ef27c0f4b98a
  13. get_obj_from_str helper · level 1 · ran · our draft was wrong · 221b2d116fdf1032

harvested Assembly for Multinomial Diffusion (graph method: Multinomial Diffusion). 11 bricks; individually on a fixture: 10 ran (of which 1 constructed an object rather than computing a result; 8 with no instrument failure: 5 honoured, 2 violated, 1 with no contract checked; 2 where Syntology's instrument failed) · 1 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. MultinomialDiffusion entry · level 1 · ran · 4268dcf6913cd21f
  2. cosine_beta_schedule scheduler · level 1 · ran · honoured contract · c92a02b48d8fb1c4
  3. default helper · level 1 · ran · violated contract · fbf9ec7be545688e
  4. extract helper · level 0 · unverified · 09c8479d9a5b3e06
  5. index_to_log_onehot data_transform · level 1 · ran; the contract check could not run · d70582d84f29d56e
  6. log_1_min_a helper · level 1 · ran · honoured contract · 479771b269fc5883
  7. log_add_exp helper · level 1 · ran · honoured contract · 032c0dece6594020
  8. log_categorical loss · level 1 · ran · honoured contract · f4b18dd8f67674cb
  9. log_onehot_to_index data_transform · level 1 · ran · honoured contract · 5def5f79a41cfc04
  10. sum_except_batch helper · level 1 · ran · our draft was wrong · 38d8ef27c0f4b98a
  11. exists helper · level 1 · ran · violated contract · 608e364a9d2376a3

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2102.05379") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

DenoisingImage SegmentationLanguage ModellingSegmentationSemantic Segmentation

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