SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2103.15692 · ICLR · 2021

Self-Constructing Neural Networks Through Random Mutation

Samuel Schmidgall

arXiv · PDF · Open in the Atlas

Code that ran

4 of 5 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; 0 honoured the contract we drafted, 2 violated it and 0 have no contract verdict; 2 failures of our instrument; 1 no recorded run); 2 of them were also run on the host, not in the jail, by the peer-agreement lane that set their level, 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/2103.15692.json · badge: /api/ran/2103.15692.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
samuelschmidgall/randommutationsearch canonical (classifier label)
classifier basis: classifier_2025
3 ran · 0 no recorded run
SamuelSchmidgall/RandomMutationSearch marked official in Papers with Code
Papers with Code archive
1 ran · 1 no recorded run
FunctionStatusWhere it lives
sigmoid Ran
ran, violated the contract we drafted
samuelschmidgall/randommutationsearch/random_mutation_search.py
code served (permissive licence) · get_code("2f53c83eaf660d6a")
How it ran

Ran in Syntology's jail on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned ndarray, ndim 1, dtype family float, shape (4,) as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-04T23:07:04Z). 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

# The sigmoid function takes a single parameter 'x'
# which is used in arithmetic operations (division, addition, tanh)
# So x should be a numpy array or scalar that supports these operations

x = np.array([0.0, 1.0, -1.0, 2.0])

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 (4,) — shape (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, violated the contract we drafted
samuelschmidgall/randommutationsearch/random_mutation_search.py
code served (permissive licence) · get_code("7f9b15275ca5759c")
How it ran

Ran in Syntology's jail on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned ndarray, ndim 1, dtype family float, shape (4,) as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-05T00:37:48Z). 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

# The softmax function takes a single parameter 'x' which should be a numpy array
# It performs exponential normalization on the input

# Create a small test input - a 1D array
x = np.array([1.0, 2.0, 3.0, 4.0])

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
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 (4,) — shape (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
_check_remove_node_out Ran
ran; our contract draft was wrong, not the code
SamuelSchmidgall/RandomMutationSearch/random_mutation_search.py
code served (permissive licence) · get_code("6107b6171e38ac0a")
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, ndim 1, 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.

def _check_remove_node_out(sub_elite, node, outg, incm):
    """ Recursively remove connecting incoming
       and outgoing nodes if ~either~ outgoing or incoming empty """
    remove_count = 0
    if len(incm) > 0:
        for _node in incm:
            sub_elite['nodes'][_node]['outgoing'].remove(node)
            remove_count += 1
            if len(sub_elite['nodes'][_node]['outgoing']) == 0 or len(sub_elite['nodes'][_node]['incoming']) == 0:
                if not (_node in sub_elite['input'] or _node in sub_elite['output']):
                    sub_elite, remove_cp = _check_remove_node_out(sub_elite, _node, sub_elite['nodes'][_node]['outgoing'], sub_elite['nodes'][_node]['incoming'])
                    del sub_elite['nodes'][_node]
                    remove_cp += remove_count
    if len(outg) > 0:
        for _node in outg:
            sub_elite['nodes'][_node]['incoming'].remove(node)
            remove_count += 1
            if len(sub_elite['nodes'][_node]['outgoing']) == 0 or len(sub_elite['nodes'][_node]['incoming']) == 0:
                if not (_node in sub_elite['input'] or _node in sub_elite['output']):
                    sub_elite, remove_cp = _check_remove_node_out(sub_elite, _node, sub_elite['nodes'][_node]['outgoing'], sub_elite['nodes'][_node]['incoming'])
                    remove_count += 1
                    del sub_elite['nodes'][_node]
                    remove_cp += remove_count
    return (sub_elite, remove_count)


# Build a minimal but valid structure for sub_elite
sub_elite_dict = {
    'nodes': {
        'node_1': {
            'incoming': ['node_0'],
            'outgoing': ['node_2']
        },
        'node_2': {
            'incoming': ['node_1'],
            'outgoing': []
        },
        'node_0': {
            'incoming': [],
            'outgoing': ['node_1']
        }
    },
    'input': ['node_0'],
    'output': ['node_2']
}

# Parameters:
# - sub_elite: the dict structure above
# - node: a node identifier to process
# - outg: list of outgoing node identifiers
# - incm: list of incoming node identifiers

ARGS = {
    'sub_elite': sub_elite_dict,
    'node': 'node_1',
    'outg': ['node_2'],
    'incm': ['node_0']
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
remove_weight Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
samuelschmidgall/randommutationsearch/random_mutation_search.py
code served (permissive licence) · get_code("a0348ca4552843d4")
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) and returned tuple, ndim 1, 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 torch
from copy import deepcopy

torch.manual_seed(0)

# Build a minimal sub_elite structure that matches what remove_weight expects
sub_elite = {
    'nodes': {
        0: {
            'outgoing': [1, 2],
            'incoming': [],
            'weights': {1: torch.tensor(0.5), 2: torch.tensor(0.3)}
        },
        1: {
            'outgoing': [3],
            'incoming': [0],
            'weights': {3: torch.tensor(0.7)}
        },
        2: {
            'outgoing': [3],
            'incoming': [0],
            'weights': {3: torch.tensor(0.2)}
        },
        3: {
            'outgoing': [],
            'incoming': [1, 2],
            'weights': {}
        }
    },
    'input': [0],
    'output': [3],
    'weight_optim': True
}

# Connection to remove: from node 0 to node 1
conn = (0, 1)

# Decay type options: 'full', 'simple', 'incoming_del', 'outgoing_del'
decay_type = 'simple'

ARGS = {
    'sub_elite': deepcopy(sub_elite),
    'conn': conn,
    'decay_type': decay_type
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
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
mutate Not yet run
unverified (harvested, no recorded run)
SamuelSchmidgall/RandomMutationSearch/random_mutation_search.py
code served (permissive licence) · get_code("83bb09bb318a8a52")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 4 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
softmax 12 other papers
3 dated earlier, 9 later
1411.2738 (2014-11)
word2vec Parameter Learning Explained
1803.05337 (2018-03)
Learning to Recognize Musical Genre from Audio
1906.05394 (2019-06)
Neural Arabic Question Answering
and 9 more not named here

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 search for neural architecture is producing many of the most exciting results in artificial intelligence. It has increasingly become apparent that task-specific neural architecture plays a crucial role for effectively solving problems. This paper presents a simple method for learning neural architecture through random mutation. This method demonstrates 1) neural architecture may be learned during the agent's lifetime, 2) neural architecture may be constructed over a single lifetime without any initial connections or neurons, and 3) architectural modifications enable rapid adaptation to dynamic and novel task scenarios. Starting without any neurons or connections, this method constructs a neural architecture capable of high-performance on several tasks. The lifelong learning capabilities of this method are demonstrated in an environment without episodic resets, even learning with constantly changing morphology, limb disablement, and changing task goals all without losing locomotion capabilities.

For agents

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

get_harvested_code_for_paper("2103.15692")
get_code_for_paper("2103.15692")
have("2103.15692")

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

curl https://syntology.ai/api/ran/2103.15692.json

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

[![Syntology run record](https://syntology.ai/api/ran/2103.15692.svg)](https://syntology.ai/paper/2103.15692)

Connect an agent — have() is free.