SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2006.04647 · ICML · 2021

Neural Architecture Search without Training

Amos Storkey, Joseph Mellor, Jack Turner, Elliot Crowley

arXiv · PDF · Open in the Atlas

Code that ran

8 of 10 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 0 violated it and 4 have no contract verdict; 3 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2006.04647.json · badge: /api/ran/2006.04647.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
nlinc1905/evolutionary-reinforcement-learner community (classifier label)
classifier basis not recorded on the link
5 ran · 2 no recorded run
idstcv/ZenNAS reimplementation (classifier label)
classifier basis: classifier_2025
2 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
1 ran · 0 no recorded run
FunctionStatusWhere it lives
compute_centered_ranks Ran
ran on a synthesized input
nlinc1905/evolutionary-reinforcement-learner/utils.py
code served (permissive licence) · get_code("693ed9995a4ce7ea")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family float, shape (6,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a simple 1D numpy array of fitness/reward values
# This represents the fitness scores for a generation of candidate solutions
x = np.array([2.5, 1.2, 8.3, 4.1, 0.9, 6.7], dtype=float)

ARGS = {
    'x': x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family float, shape (6,) — shape (6,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compute_nas_score Ran
ran; our contract draft was wrong, not the code
idstcv/ZenNAS/ZeroShotProxy/compute_zen_score.py
pointer only (licence: NONE) · get_code("866a541b53979f13")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, idstcv/ZenNAS when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

get_flop_loss Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from VITA-Group/architecture_convergence
pointer only · get_code("5aa7497aa00af169")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

network_weight_gaussian_init Ran
ran; our contract draft was wrong, not the code
idstcv/ZenNAS/ZeroShotProxy/compute_zen_score.py
pointer only (licence: NONE) · get_code("28d3ae653452ecdf")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, idstcv/ZenNAS when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

quadratic_fxn_fitness Ran
ran on a synthesized input
nlinc1905/evolutionary-reinforcement-learner/reward_functions/fitness_functions.py
code served (permissive licence) · get_code("acf41416a7469905")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned float64, ndim 0, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

ARGS = {
    'params': np.array([0.5, 2.0, -1.5])
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float64, ndim 0, dtype family float — no output shape is stated: observe_signature.py recorded float ndim 0, which disagrees with the fixture run
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
relu Ran
ran, honoured the contract we drafted
nlinc1905/evolutionary-reinforcement-learner/models/mlp.py
code served (permissive licence) · get_code("0bd581e0afb4c04a")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small 2D numpy array for testing the relu function
# Using shape (2, 4) - batch size 2, feature dimension 4
x = np.array([
    [-1.0, 2.0, -3.0, 4.0],
    [5.0, -2.0, 0.0, -0.5]
], dtype=np.float32)

ARGS = {
    'x': x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
softmax Ran
ran on a synthesized input
nlinc1905/evolutionary-reinforcement-learner/models/mlp.py
code served (permissive licence) · get_code("2374481500fc132c")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small 2D numpy array representing logits from a neural network
# Shape: (batch_size=2, num_actions=4)
# This represents output layer activations before softmax
a = np.array([
    [1.0, 2.0, 3.0, 0.5],
    [2.5, 1.5, 0.5, 3.5]
], dtype=np.float32)

ARGS = {
    'a': a
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
train_function_optimizing_agent Ran
ran on a synthesized input
nlinc1905/evolutionary-reinforcement-learner/examples.py
code served (permissive licence) · get_code("248ec46cc977d5ae")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch
import torch.nn as nn

torch.manual_seed(0)

class MockOptimizer:
    """Mock optimizer that implements the evolve() method."""
    def __init__(self):
        self.generations = 5
        
    def evolve(self):
        # Return (optimal_params, generational_fitness)
        optimal_params = np.array([0.1, 0.2, 0.3, 0.4])
        generational_fitness = np.array([1.0, 2.0, 3.0, 4.0, 5.0])
        return optimal_params, generational_fitness
    
    def __str__(self):
        return "<MockOptimizer object at 0x12345>"

ARGS = {
    "optimizer": MockOptimizer(),
    "plot_learning_curve": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_mean_and_standardized_rewards Not yet run
unverified (harvested, no recorded run)
nlinc1905/evolutionary-reinforcement-learner/utils.py
code served (permissive licence) · get_code("3817dba1742caf11")
How it ran

no invocation recorded.

mutate Not yet run
unverified (harvested, no recorded run)
nlinc1905/evolutionary-reinforcement-learner/utils.py
code served (permissive licence) · get_code("91e7124a049d07e9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 4 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 4 later) and 6 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
relu 5 other papers
2 dated earlier, 3 later
1907.10599 (2019-07)
A Fine-Grained Spectral Perspective on Neural Networks
2003.04884 (2020-03)
Cryptanalytic Extraction of Neural Network Models
2210.02419 (2022-10)
Boundary-Aware Uncertainty for Feature Attribution Explainers
and 2 more not named here
compute_nas_score 1 other paper
1 later
2102.01063 (2021-02)
Zen-NAS: A Zero-Shot NAS for High-Performance Image Recognition
get_flop_loss 1 other paper
1 later
2205.05662 (2022-05)
Deep Architecture Connectivity Matters for Its Convergence: A Fine-Grained Analysis
network_weight_gaussian_init 1 other paper
1 later
2102.01063 (2021-02)
Zen-NAS: A Zero-Shot NAS for High-Performance Image Recognition

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

The time and effort involved in hand-designing deep neural networks is immense. This has prompted the development of Neural Architecture Search (NAS) techniques to automate this design. However, NAS algorithms tend to be slow and expensive; they need to train vast numbers of candidate networks to inform the search process. This could be alleviated if we could partially predict a network's trained accuracy from its initial state. In this work, we examine the overlap of activations between datapoints in untrained networks and motivate how this can give a measure which is usefully indicative of a network's trained performance. We incorporate this measure into a simple algorithm that allows us to search for powerful networks without any training in a matter of seconds on a single GPU, and verify its effectiveness on NAS-Bench-101, NAS-Bench-201, NATS-Bench, and Network Design Spaces. Our approach can be readily combined with more expensive search methods; we examine a simple adaptation of regularised evolutionary search. Code for reproducing our experiments is available at https://github.com/ BayesWatch/nas-without-training.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2006.04647")
get_code_for_paper("2006.04647")
have("2006.04647")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2006.04647.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2006.04647.svg)](https://syntology.ai/paper/2006.04647)

Connect an agent — have() is free.