Papers › Neural Additive Models: Interpretable Machine Learning with Neural Nets

Neural Additive Models: Interpretable Machine Learning with Neural Nets

29 Apr 2020NeurIPS 2021 12arXiv:2004.13912archive 2025-07-28

Rishabh Agarwal, Levi Melnick, Nicholas Frosst, Xuezhou Zhang, Ben Lengerich, Rich Caruana, Geoffrey Hinton

Deep neural networks (DNNs) are powerful black-box predictors that have achieved impressive performance on a wide variety of tasks. However, their accuracy comes at the cost of intelligibility: it is usually unclear how they make their decisions. This hinders their applicability to high stakes decision-making domains such as healthcare. We propose Neural Additive Models (NAMs) which combine some of the expressivity of DNNs with the inherent intelligibility of generalized additive models. NAMs learn a linear combination of neural networks that each attend to a single input feature. These networks are trained jointly and can learn arbitrarily complex relationships between their input feature and the output. Our experiments on regression and classification datasets show that NAMs are more accurate than widely used intelligible models such as logistic regression and shallow decision trees. They perform similarly to existing state-of-the-art generalized additive models in accuracy, but are more flexible because they are based on neural nets instead of boosted trees. To demonstrate this, we show how NAMs can be used for multitask learning on synthetic data and on the COMPAS recidivism data due to their composability, and demonstrate that the differentiability of NAMs allows them to train more complex interpretable models for COVID-19.

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.

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

Code

Syntology 33 code samples harvested from 8 repositories linked to this paper: 26 ran (of which 14 constructed an object rather than computing a result; 26 with no instrument failure: 5 honoured, 0 violated, 21 with no contract checked; 0 where Syntology's instrument failed) · 7 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code; here there are none.

By repository: official repository: 2 samples from 1 repository, 1 ran (1 with no instrument failure; 0 where Syntology's instrument failed); community (archive-listed): 31 samples from 7 repositories, 25 ran (25 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 between 2026-09-22 and 2026-09-24 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

lemeln/nam officialpytorchSyntology: 2 samples, 1 ran (0 instrument-failed) · 1 unverified report
AmrMKayid/nam mentioned on GitHubpytorchSyntology: 5 samples, 5 ran (0 instrument-failed) · 0 unverified report
Habush/nam_jax mentioned on GitHubjaxSyntology: 5 samples, 4 ran (0 instrument-failed) · 1 unverified report
fortuinlab/la-nam mentioned on GitHubpytorchMITSyntology: 4 samples, 3 ran (0 instrument-failed) · 1 unverified report
kherud/neural-additive-models-pt mentioned on GitHubpytorchSyntology: 6 samples, 5 ran (0 instrument-failed) · 1 unverified report
nickfrosst/neural_additive_models mentioned on GitHubtfSyntology: 6 samples, 5 ran (0 instrument-failed) · 1 unverified report
saipradeep-peri/Neural_Additive_Models_Walkthrough mentioned on GitHubpytorchSyntology: 2 samples, 2 ran (0 instrument-failed) · 0 unverified report
google-research/google-research tfSyntology: 3 samples, 1 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

33 samples harvested: 26 ran (of which 14 constructed an object rather than computing a result; 26 with no instrument failure: 5 honoured, 0 violated, 21 with no contract checked; 0 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 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.

5ran · honoured contract
21ran
7unverified

Licence: 14 of the 33 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 8 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”. For 6 samples that ran, the graph records a signature this page does not show: for a class the fixture built without calling, the harness called the object with inputs it made up itself, so the recorded signature is the harness's probe, not the fixture's arguments. “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.

lemeln/nam official repository 2 samples, 1 ran (0 instrument-failed) · 1 unverified
FeatureNN nam/models/nam.py ran MIT (permissive) · b0e0374e9293aa60 · report fixture: signature inputs:Tensor/2/float · returned Tensor · level recorded 2026-09-24
NAM nam/models/nam.py unverified MIT (permissive) · 8ef351ea39fd6b03 · report
AmrMKayid/nam community (archive-listed) 5 samples, 5 ran (0 instrument-failed) · 0 unverified
ExU nam/models/nam.py ran MIT (permissive) · 4906a2527041456e · report fixture: signature inputs:Tensor/2/float, n:int/0/int · returned Tensor · level recorded 2026-09-24
FeatureNN nam/models/nam.py ran fingerprinted MIT (permissive) · dd231b4d8f71426e · report fixture: returned FeatureNN · level recorded 2026-09-22
LinReLU nam/models/nam.py ran MIT (permissive) · 80fc1bfd32c3dd88 · report fixture: returned LinReLU · level recorded 2026-09-22
Model in 227 papers nam/models/nam.py ran MIT (permissive) · d3617b2e6bdec534 · report fixture: returned Model · level recorded 2026-09-22
NAM nam/models/nam.py ran fingerprinted MIT (permissive) · a38b31cb2b4023e6 · report fixture: returned NAM · level recorded 2026-09-22
Habush/nam_jax community (archive-listed) 5 samples, 4 ran (0 instrument-failed) · 1 unverified
NAM models.py ran Apache-2.0 (permissive) · 4a9811b70594bfe1 · report fixture: returned NAM · level recorded 2026-09-22
TrainingState models.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 22b5675b6955ad08 · report fixture: returned TrainingState · level recorded 2026-09-22
exu models.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · efd59dcfa39ea264 · report fixture: signature x:ArrayImpl/2/float, weight:ArrayImpl/2/float, bias:ArrayImpl/1/float · returned ArrayImpl · level recorded 2026-09-22
relu in 58 papers
census: 8 implementations from 11 papers, 4 distinct outputs across 2 buckets, one shared input per bucket
models.py ran · honoured contract fingerprinted Apache-2.0 (permissive) · 454cfa9b8b9aed47 · report fixture: signature x:ArrayImpl/2/float, weight:ArrayImpl/2/float, bias:ArrayImpl/1/float · returned ArrayImpl · level recorded 2026-09-22
FeatureNet in 7 papers models.py unverified Apache-2.0 (permissive) · 8deb3a12b331286f · report
fortuinlab/la-nam community (archive-listed) 4 samples, 3 ran (0 instrument-failed) · 1 unverified
Categorical in 6 papers src/laplace_skorch/modules.py ran · metamorphic tier: invariant fingerprinted MIT (permissive) · 2c6a77de4277d87c · report fixture: signature input:Tensor/2/int · returned Tensor · level recorded 2026-09-22
Mixed src/laplace_skorch/modules.py ran fingerprinted MIT (permissive) · a3bc285602224f16 · report fixture: signature input:Tensor/2/float · returned Tensor · level recorded 2026-09-24
Numerical src/laplace_skorch/modules.py ran · metamorphic tier: invariant MIT (permissive) · 929c1f2bcaed27ca · report fixture: signature input:Tensor/2/float · returned Tensor · level recorded 2026-09-22
Router in 11 papers src/laplace_skorch/modules.py unverified MIT (permissive) · 9e7fc6cfe97a24f2 · report
kherud/neural-additive-models-pt community (archive-listed) 6 samples, 5 ran (0 instrument-failed) · 1 unverified
ActivationLayer in 7 papers nam/model.py ran no licence file found · pointer only · 51c75dde5a5a0f26 · report fixture: returned ActivationLayer · level recorded 2026-09-22
ExULayer nam/model.py ran no licence file found · pointer only · 2d8839407b350dc9 · report fixture: returned ExULayer · level recorded 2026-09-22
FeatureNN nam/model.py ran no licence file found · pointer only · 8c55bd802f9906ca · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded 2026-09-24
NeuralAdditiveModel nam/model.py ran fingerprinted no licence file found · pointer only · cae4311ff1f77d90 · report fixture: signature x:Tensor/2/float · returned tuple · level recorded 2026-09-24
ReLULayer nam/model.py ran no licence file found · pointer only · 108769e8a1a48493 · report fixture: returned ReLULayer · level recorded 2026-09-22
truncated_normal_ in 8 papers nam/model.py unverified no licence file found · pointer only · 9ea962308449a4bb · report
nickfrosst/neural_additive_models community (archive-listed) 6 samples, 5 ran (0 instrument-failed) · 1 unverified
ActivationLayer in 7 papers models.py ran no licence file found · pointer only · 546b56723399f9fe · report fixture: returned ActivationLayer · level recorded 2026-09-22
FeatureNN models.py ran no licence file found · pointer only · 053cdbd7b148f0cc · report fixture: returned FeatureNN · level recorded 2026-09-22
exu models.py ran · honoured contract fingerprinted no licence file found · pointer only · 8a597de16bdccb7a · report fixture: signature x:EagerTensor/2/float, weight:EagerTensor/2/float, bias:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-22
relu in 58 papers
census: 8 implementations from 11 papers, 4 distinct outputs across 2 buckets, one shared input per bucket
models.py ran · honoured contract fingerprinted no licence file found · pointer only · 207a3a66ecdc0f2f · report fixture: signature x:EagerTensor/2/float, weight:EagerTensor/2/float, bias:EagerTensor/2/float · returned EagerTensor · level recorded 2026-09-22
relu_n models.py ran · honoured contract no licence file found · pointer only · 2c40ca3eaec98f16 · report fixture: signature x:EagerTensor/2/float, n:int/0/int · returned SymbolicTensor · level recorded 2026-09-22
NAM models.py unverified no licence file found · pointer only · 2c18cadeb8c1c012 · report
saipradeep-peri/Neural_Additive_Models_Walkthrough community (archive-listed) 2 samples, 2 ran (0 instrument-failed) · 0 unverified
FeatureNN nam/models/nam.py ran fingerprinted no licence file found · pointer only · 243e71c28636308b · report fixture: returned FeatureNN · level recorded 2026-09-22
NAM nam/models/nam.py ran no licence file found · pointer only · 9c956ecf224f6108 · report fixture: returned NAM · level recorded 2026-09-22
google-research/google-research community (archive-listed) 3 samples, 1 ran (0 instrument-failed) · 2 unverified
FeatureNN neural_additive_models/models.py ran Apache-2.0 (permissive) · 7a5f01663e901790 · report fixture: returned FeatureNN · level recorded 2026-09-22
ActivationLayer in 7 papers neural_additive_models/models.py unverified Apache-2.0 (permissive) · 9e20b229f2c0a6a3 · report
NAM neural_additive_models/models.py unverified Apache-2.0 (permissive) · d362a52791a5c0e9 · 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. 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. 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.

8 assemblies on this paper: 8 harvested, 0 generated, listed in 1 groups (siblings lifted in one batch form one group).

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

harvested Assembly for Neural Additive Models (graph method: Neural Additive Models). 5 bricks; individually on a fixture: 4 ran (of which 2 constructed an object rather than computing a result; 4 with no instrument failure: 2 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified. The assembly has not been run end-to-end.

one of 8 siblings

  1. NAM entry · level 1 · ran · 4a9811b70594bfe1
  2. FeatureNet model · level 0 · unverified · 8deb3a12b331286f
  3. TrainingState helper · level 1 · ran · 22b5675b6955ad08
  4. exu helper · level 1 · ran · honoured contract · efd59dcfa39ea264
  5. relu helper · level 1 · ran · honoured contract · 454cfa9b8b9aed47

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

one of 8 siblings

  1. NeuralAdditiveModel entry · level 1 · ran · cae4311ff1f77d90
  2. ActivationLayer model · level 1 · ran · 51c75dde5a5a0f26
  3. ExULayer model · level 1 · ran · 2d8839407b350dc9
  4. FeatureNN model · level 1 · ran · 8c55bd802f9906ca
  5. ReLULayer model · level 1 · ran · 108769e8a1a48493
  6. truncated_normal_ helper · level 0 · unverified · 9ea962308449a4bb

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

one of 8 siblings

  1. NAM entry · level 1 · ran · a38b31cb2b4023e6
  2. FeatureNN model · level 1 · ran · dd231b4d8f71426e
  3. Model helper · level 1 · ran · d3617b2e6bdec534
  4. ExU helper · level 1 · ran · 4906a2527041456e
  5. LinReLU helper · level 1 · ran · 80fc1bfd32c3dd88

harvested Assembly for Neural Additive Models (graph method: Neural Additive Models). 6 bricks; individually on a fixture: 5 ran (of which 2 constructed an object rather than computing a result; 5 with no instrument failure: 3 honoured, 0 violated, 2 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified. The assembly has not been run end-to-end.

one of 8 siblings

  1. NAM entry · level 0 · unverified · 2c18cadeb8c1c012
  2. FeatureNN model · level 1 · ran · 053cdbd7b148f0cc
  3. ActivationLayer model · level 1 · ran · 546b56723399f9fe
  4. exu model · level 1 · ran · honoured contract · 8a597de16bdccb7a
  5. relu model · level 1 · ran · honoured contract · 207a3a66ecdc0f2f
  6. relu_n model · level 1 · ran · honoured contract · 2c40ca3eaec98f16

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

one of 8 siblings

  1. NAM entry · level 1 · ran · 9c956ecf224f6108
  2. FeatureNN model · level 1 · ran · 243e71c28636308b
  3. Model helper · level 1 · ran · d3617b2e6bdec534
  4. ExU helper · level 1 · ran · 4906a2527041456e
  5. LinReLU helper · level 1 · ran · 80fc1bfd32c3dd88

harvested Assembly for Neural Additive Models (graph method: Neural Additive Models). 6 bricks; individually on a fixture: 4 ran (of which 1 constructed an object rather than computing a result; 4 with no instrument failure: 3 honoured, 0 violated, 1 with no contract checked; 0 where Syntology's instrument failed) · 2 unverified. The assembly has not been run end-to-end.

one of 8 siblings

  1. NAM entry · level 0 · unverified · d362a52791a5c0e9
  2. FeatureNN model · level 1 · ran · 7a5f01663e901790
  3. ActivationLayer model · level 0 · unverified · 9e20b229f2c0a6a3
  4. exu model · level 1 · ran · honoured contract · 8a597de16bdccb7a
  5. relu model · level 1 · ran · honoured contract · 207a3a66ecdc0f2f
  6. relu_n model · level 1 · ran · honoured contract · 2c40ca3eaec98f16

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

one of 8 siblings

  1. Router entry · level 0 · unverified · 9e7fc6cfe97a24f2
  2. Categorical model · level 1 · ran · 2c6a77de4277d87c
  3. Mixed model · level 1 · ran · a3bc285602224f16
  4. Numerical model · level 1 · ran · 929c1f2bcaed27ca

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

one of 8 siblings

  1. NAM entry · level 0 · unverified · 8ef351ea39fd6b03
  2. FeatureNN model · level 1 · ran · b0e0374e9293aa60
  3. ExU model · level 1 · ran · 4906a2527041456e
  4. LinReLU model · level 1 · ran · 80fc1bfd32c3dd88

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="2004.13912") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Additive modelsBIG-bench Machine LearningDecision MakingInterpretable Machine Learningregression

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

Introduced by this paper: NAM

Logistic RegressionNAM

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