Papers › Geometry-Aware Gradient Algorithms for Neural Architecture Search

Geometry-Aware Gradient Algorithms for Neural Architecture Search

16 Apr 2020ICLR 2021 1arXiv:2004.07802archive 2025-07-28

Liam Li, Mikhail Khodak, Maria-Florina Balcan, Ameet Talwalkar

Recent state-of-the-art methods for neural architecture search (NAS) exploit gradient-based optimization by relaxing the problem into continuous optimization over architectures and shared-weights, a noisy process that remains poorly understood. We argue for the study of single-level empirical risk minimization to understand NAS with weight-sharing, reducing the design of NAS methods to devising optimizers and regularizers that can quickly obtain high-quality solutions to this problem. Invoking the theory of mirror descent, we present a geometry-aware framework that exploits the underlying structure of this optimization to return sparse architectural parameters, leading to simple yet novel algorithms that enjoy fast convergence guarantees and achieve state-of-the-art accuracy on the latest NAS benchmarks in computer vision. Notably, we exceed the best published results for both CIFAR and ImageNet on both the DARTS search space and NAS-Bench201; on the latter we achieve near-oracle-optimal performance on CIFAR-10 and CIFAR-100. Together, our theory and experiments demonstrate a principled way to co-design optimizers and continuous relaxations of discrete NAS search spaces.

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 GEOMETRY-AWARE GRADIENT ALGORITHMS, 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). No rung compares this code against the authors' implementation. For agents: get_reference_implementation(name='GEOMETRY-AWARE GRADIENT ALGORITHMS') (how to connect); the free check have('2004.07802') needs no token.

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

Code

Syntology 15 code samples harvested from 3 repositories linked to this paper: 12 ran (of which 9 constructed an object rather than computing a result; 11 with no instrument failure: 1 honoured, 0 violated, 10 with no contract checked; 1 where Syntology's instrument failed) · 3 unverified. Where Syntology's instrument failed: 1 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 5 samples from 1 repository, 3 ran (2 with no instrument failure; 1 where Syntology's instrument failed); found in paper text by Syntology: 10 samples from 2 repositories, 9 ran (9 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.

liamcli/gaea_release officialmentioned in papermentioned on GitHubpytorchSyntology: 5 samples, 3 ran (1 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

15 samples harvested: 12 ran (of which 9 constructed an object rather than computing a result; 11 with no instrument failure: 1 honoured, 0 violated, 10 with no contract checked; 1 where Syntology's instrument failed) · 3 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 · our draft was wrong
10ran
3unverified

Licence: 5 of the 15 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 3 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 5 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.

liamcli/gaea_release official repository 5 samples, 3 ran (1 instrument-failed) · 2 unverified
AdaptiveLR cnn/architect/architect_edarts.py ran no licence file found · pointer only · 0ae6d8b3f048531d · report fixture: returned AdaptiveLR · level recorded 2026-09-22
History cnn/architect/architect_edarts.py ran no licence file found · pointer only · 2b798f4102f93552 · report fixture: returned History · level recorded 2026-09-22
normalize in 638 papers
census: 70 implementations from 82 papers, 51 distinct outputs across 11 buckets, one shared input per bucket this sample: bucket 1 of 11, arg 1: rank 2, kind float, dtype float32
cnn/architect/architect_edarts.py ran · our draft was wrong fingerprinted no licence file found · pointer only · 35516df482cccb4b · report fixture: signature x:Tensor/2/float, dim:int/0/int, min_v:float/0/float · returned Tensor · level recorded 2026-09-22
Architect in 5 papers cnn/architect/architect_edarts.py unverified no licence file found · pointer only · e70f49be6527cff9 · report
ArchitectEDARTS cnn/architect/architect_edarts.py unverified no licence file found · pointer only · 411c70cf0de59d37 · report
automl/nasbench-1shot1 found in paper text by Syntology 2 samples, 2 ran (0 instrument-failed) · 0 unverified
Architect in 5 papers optimizers/darts/architect.py ran Apache-2.0 (permissive) · b0d6e9e65ecc665d · report fixture: returned Architect · level recorded 2026-09-22
_concat optimizers/darts/architect.py ran · honoured contract Apache-2.0 (permissive) · acb824e0877d8cbd · report fixture: signature xs:list/1/Tensor · returned Tensor · level recorded 2026-09-22
NivNayman/XNAS found in paper text by Syntology 8 samples, 7 ran (0 instrument-failed) · 1 unverified
DilConv model.py ran · metamorphic tier: deterministic fingerprinted Apache-2.0 (permissive) · 132f546bb7ef86ba · report fixture: returned DilConv · level recorded 2026-09-22
FactorizedReduce model.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · 5acfdb2e37fd4c79 · report fixture: returned FactorizedReduce · level recorded 2026-09-22
Identity in 49 papers
census: 10 implementations from 13 papers, 4 distinct outputs across 4 buckets, one shared input per bucket this sample: bucket 1 of 4, arg 1: rank 2, kind float, dtype float32
model.py ran · metamorphic tier: invariant fingerprinted Apache-2.0 (permissive) · c6c6288219c9d532 · report fixture: signature x:Tensor/2/float · returned Tensor · level recorded 2026-09-22
ReLUConvBN model.py ran · metamorphic tier: deterministic fingerprinted Apache-2.0 (permissive) · 2b614488c8cb14d3 · report fixture: returned ReLUConvBN · level recorded 2026-09-22
SepConv in 8 papers model.py ran · metamorphic tier: deterministic fingerprinted Apache-2.0 (permissive) · 290cc06f55a8d0ed · report fixture: returned SepConv · level recorded 2026-09-22
SepConvRes model.py ran · metamorphic tier: deterministic Apache-2.0 (permissive) · c585344417dc92cc · report fixture: returned SepConvRes · level recorded 2026-09-22
Zero in 5 papers model.py ran · metamorphic tier: deterministic fingerprinted Apache-2.0 (permissive) · bc41443a9046202e · report fixture: returned Zero · level recorded 2026-09-22
Cell in 12 papers model.py unverified Apache-2.0 (permissive) · f4c89acbb418f526 · 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.

4 assemblies on this paper: 3 harvested, 1 generated, listed in 2 groups (siblings lifted in one batch form one group).

generated Assembly for GEOMETRY-AWARE GRADIENT ALGORITHMS (graph method: GEOMETRY-AWARE GRADIENT ALGORITHMS), generated from the paper text, not the repository's code. 1 brick; individually on a fixture: 1 ran (1 with no instrument failure: 0 honoured, 1 violated, 0 with no contract checked; 0 where Syntology's instrument failed) · 0 unverified. Ran end-to-end against an independent peer (level 3; 5 property tests passed on both).

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

  1. mirror_descent_architecture_update entry · level 3 · ran · violated contract · 274cd11f6e968aa5dac5dcb1b14d7bf875c9f46450d3f28ab2b4aa7d86c95bfd

3 sibling assemblies for GEOMETRY-AWARE GRADIENT ALGORITHMS (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for GEOMETRY-AWARE GRADIENT ALGORITHMS (graph method: GEOMETRY-AWARE GRADIENT ALGORITHMS). 8 bricks; individually on a fixture: 7 ran (of which 6 constructed an object rather than computing a result; 7 with no instrument failure: 0 honoured, 0 violated, 7 with no contract checked; 0 where Syntology's instrument failed) · 1 unverified. The assembly has not been run end-to-end.

one of 3 siblings

  1. Cell entry · level 0 · unverified · f4c89acbb418f526
  2. FactorizedReduce model · level 1 · ran · 5acfdb2e37fd4c79
  3. Identity model · level 1 · ran · c6c6288219c9d532
  4. ReLUConvBN model · level 1 · ran · 2b614488c8cb14d3
  5. DilConv model · level 1 · ran · 132f546bb7ef86ba
  6. SepConv model · level 1 · ran · 290cc06f55a8d0ed
  7. SepConvRes model · level 1 · ran · c585344417dc92cc
  8. Zero model · level 1 · ran · bc41443a9046202e

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

one of 3 siblings

  1. ArchitectEDARTS entry · level 0 · unverified · 411c70cf0de59d37
  2. AdaptiveLR helper · level 1 · ran · 0ae6d8b3f048531d
  3. Architect helper · level 0 · unverified · e70f49be6527cff9
  4. History helper · level 1 · ran · 2b798f4102f93552
  5. normalize helper · level 1 · ran · our draft was wrong · 35516df482cccb4b

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

one of 3 siblings

  1. Architect entry · level 1 · ran · b0d6e9e65ecc665d
  2. _concat helper · level 1 · ran · honoured contract · acb824e0877d8cbd

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="2004.07802") returns these assemblies with each brick's role and wiring (how to connect).

Tasks

Neural Architecture Search

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Neural Architecture Search ImageNet GAEA PC-DARTS Params 5.6 #97 of 135 Archive leaderboard report
Neural Architecture Search ImageNet GAEA PC-DARTS Top-1 Error Rate 24 #97 of 135 Archive leaderboard report
Neural Architecture Search NAS-Bench-201, CIFAR-10 GAEA DARTS (ERM) Accuracy (Test) 94.1 #16 of 37 Archive leaderboard report
Neural Architecture Search NAS-Bench-201, CIFAR-100 GAEA DARTS (ERM) Accuracy (Test) 73.43 #9 of 40 Archive leaderboard report
Neural Architecture Search NAS-Bench-201, ImageNet-16-120 GAEA DARTS (ERM) Accuracy (Test) 46.36 #10 of 49 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

DARTSLSTMSigmoid ActivationSoftmaxTanh Activation

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