SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1710.05941 · 2017

SEARCHING FOR ACTIVATION FUNCTIONS

Barret Zoph, Prajit Ramachandran, Quoc Le Google Brain

arXiv · PDF · Open in the Atlas

Code that ran

13 of 20 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 1 honoured the contract we drafted, 2 violated it and 1 has no contract verdict; 9 failures of our instrument; 7 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/1710.05941.json · badge: /api/ran/1710.05941.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
WessZumino/expectation_propagation community (classifier label)
classifier basis not recorded on the link
5 ran · 3 no recorded run
Neoanarika/Searching-for-activation-functions community (classifier label)
classifier basis not recorded on the link
3 ran · 3 no recorded run
islaterm/you-activated-my-neuron community
Papers with Code archive
2 ran · 0 no recorded run
WarClans612/deep_learning_final community (classifier label)
classifier basis: plan_row
1 ran · 1 no recorded run
tyunist/memory_efficient_mish_swish community
Papers with Code archive
1 ran · 0 no recorded run
WessZumino/meanfield-theory-of-activation-functions community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
Swish_func Ran
ran on a synthesized input · constructed an object; no result was computed
tyunist/memory_efficient_mish_swish/swish.py
pointer only (licence: NONE) · get_code("637511ceeecf1796")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, tyunist/memory_efficient_mish_swish 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.

act Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WessZumino/meanfield-theory-of-activation-functions/TF/esp_tf_utils.py
code served (permissive licence) · get_code("df1af39a35cafa81")
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 EagerTensor, 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 tensorflow as tf
import numpy as np

# Create a small tensor for h
h = tf.constant([[1.0, 2.0, 3.0, 4.0],
                 [0.5, 1.5, 2.5, 3.5]], dtype=tf.float32)

# Create a scalar beta value
beta = tf.constant(0.5, dtype=tf.float32)

# Create activation function name
activation = 'esp'

ARGS = {
    'h': h,
    'beta': beta,
    'activation': activation
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, 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
act Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WessZumino/expectation_propagation/Python/esp_utils.py
code served (permissive licence) · get_code("f0ec10189ae424f4")
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

# Set random seed for reproducibility
np.random.seed(0)

# Create small test data
h = np.array([[0.5, -0.3, 1.2, -2.1],
              [0.1, 0.8, -0.5, 1.5]], dtype=np.float32)

beta = 0.5

activation = 'esp'

ARGS = {
    'h': h,
    'beta': beta,
    'activation': activation
}
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
epoch_sampling Ran
ran, honoured the contract we drafted
WessZumino/expectation_propagation/TF/analysis_utils.py
code served (permissive licence) · get_code("99c30c0e30d4744e")
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 list, ndim 1, dtype family int; 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 = {
    'epochs': 1000,
    'uniform_step': 50
}
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 list, ndim 1, dtype family int — a list 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
fixed_padding Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Neoanarika/Searching-for-activation-functions/src/childnetwork.py
code served (permissive licence) · get_code("5756d6bfbcce2dfc")
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 EagerTensor, ndim 4, dtype family float, shape (2, 3, 6, 7) 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 tensorflow as tf

# Create a small tensor with batch=2, channels=3, height=4, width=5
# Using channels_first format: [batch, channels, height, width]
inputs_channels_first = tf.constant(
    [[[[1, 2, 3, 4, 5],
       [6, 7, 8, 9, 10],
       [11, 12, 13, 14, 15],
       [16, 17, 18, 19, 20]],
      [[21, 22, 23, 24, 25],
       [26, 27, 28, 29, 30],
       [31, 32, 33, 34, 35],
       [36, 37, 38, 39, 40]],
      [[41, 42, 43, 44, 45],
       [46, 47, 48, 49, 50],
       [51, 52, 53, 54, 55],
       [56, 57, 58, 59, 60]]],
     [[[61, 62, 63, 64, 65],
       [66, 67, 68, 69, 70],
       [71, 72, 73, 74, 75],
       [76, 77, 78, 79, 80]],
      [[81, 82, 83, 84, 85],
       [86, 87, 88, 89, 90],
       [91, 92, 93, 94, 95],
       [96, 97, 98, 99, 100]],
      [[101, 102, 103, 104, 105],
       [106, 107, 108, 109, 110],
       [111, 112, 113, 114, 115],
       [116, 117, 118, 119, 120]]]],
    dtype=tf.float32
)

# Create tensor with channels_last format: [batch, height, width, channels]
inputs_channels_last = tf.constant(
    [[[[1, 2, 3],
       [4, 5, 6],
       [7, 8, 9],
       [10, 11, 12],
       [13, 14, 15]],
      [[16, 17, 18],
       [19, 20, 21],
       [22, 23, 24],
       [25, 26, 27],
       [28, 29, 30]],
      [[31, 32, 33],
       [34, 35, 36],
       [37, 38, 39],
       [40, 41, 42],
       [43, 44, 45]],
      [[46, 47, 48],
       [49, 50, 51],
       [52, 53, 54],
       [55, 56, 57],
       [58, 59, 60]]],
     [[[61, 62, 63],
       [64, 65, 66],
       [67, 68, 69],
       [70, 71, 72],
       [73, 74, 75]],
      [[76, 77, 78],
       [79, 80, 81],
       [82, 83, 84],
       [85, 86, 87],
       [88, 89, 90]],
      [[91, 92, 93],
       [94, 95, 96],
       [97, 98, 99],
       [100, 101, 102],
       [103, 104, 105]],
      [[106, 107, 108],
       [109, 110, 111],
       [112, 113, 114],
       [115, 116, 117],
       [118, 119, 120]]]],
    dtype=tf.float32
)

ARGS = {
    "inputs": inputs_channels_first,
    "kernel_size": 3,
    "data_format": "channels_first"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type EagerTensor, ndim 4, dtype family float, shape (2, 3, 6, 7) — shape (2, 3, 6, 7) 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
index Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
WessZumino/expectation_propagation/TF/analysis_utils.py
code served (permissive licence) · get_code("2bb7b02beb4bcb12")
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

# The function expects hessian_epoch to be a list of Hessian matrices (2D numpy arrays)
# Each Hessian matrix should be square (eigenvalues are computed from them)
# Let's create a small example with 3 epochs, each with a 4x4 Hessian matrix

# Create some sample Hessian matrices for different epochs
np.random.seed(0)

# Epoch 1: A Hessian matrix with mixed eigenvalues (some negative, some positive)
hessian_1 = np.array([
    [2.0, 0.5, 0.1, 0.0],
    [0.5, -1.0, 0.2, 0.1],
    [0.1, 0.2, 3.0, 0.3],
    [0.0, 0.1, 0.3, -0.5]
], dtype=np.float64)

# Epoch 2: Another Hessian matrix
hessian_2 = np.array([
    [1.5, 0.3, 0.0, 0.1],
    [0.3, -2.0, 0.2, 0.0],
    [0.0, 0.2, 2.5, 0.2],
    [0.1, 0.0, 0.2, -1.5]
], dtype=np.float64)

# Epoch 3: A third Hessian matrix
hessian_3 = np.array([
    [3.0, 0.2, 0.1, 0.0],
    [0.2, 0.5, 0.0, 0.2],
    [0.1, 0.0, 2.0, 0.3],
    [0.0, 0.2, 0.3, 1.0]
], dtype=np.float64)

# Create the list of Hessian matrices for different epochs
hessian_epoch = [hessian_1, hessian_2, hessian_3]

ARGS = {
    "hessian_epoch": hessian_epoch
}
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 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
initialize_parameters Ran
ran; our contract draft was wrong, not the code
WessZumino/expectation_propagation/Python/esp_utils.py
code served (permissive licence) · get_code("1c69a8fc3aa908af")
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 dict. 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

# The function takes a list of integers representing layer dimensions
# For a simple neural network with 3 layers: input layer (5 units), hidden layer (4 units), output layer (3 units)
layers = [5, 4, 3]

ARGS = {
    "layers": layers
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
path_exists Ran
ran; our contract draft was wrong, not the code
Neoanarika/Searching-for-activation-functions/src/utils.py
code served (permissive licence) · get_code("e6085d661f8724f7")
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 str. 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 os
import tempfile
import shutil

# Create a temporary directory for testing
temp_dir = tempfile.mkdtemp()
test_path = os.path.join(temp_dir, "test_dir")

ARGS = {
    "path": test_path,
    "overwrite": False
}
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 str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
remove_dir Ran
ran, violated the contract we drafted
Neoanarika/Searching-for-activation-functions/src/utils.py
code served (permissive licence) · get_code("9e3b33e6b46657bb")
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 bool, ndim 0, dtype family float, shape () 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 os
import tempfile

# Create a temporary directory to be removed
temp_dir = tempfile.mkdtemp()

ARGS = {
    "path": temp_dir
}
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 bool, ndim 0, dtype family float, shape () — shape () 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
sig Ran
ran; our contract draft was wrong, not the code
islaterm/you-activated-my-neuron/src/neuron/activation_functions.py
pointer only (licence: NONE) · get_code("f32504dbf9150ad6")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, islaterm/you-activated-my-neuron 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.

sigmoid Ran
ran, violated the contract we drafted
WessZumino/expectation_propagation/Python/esp_utils.py
code served (permissive licence) · get_code("3b622bff5ce900a3")
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 (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

# Construct arguments for sigmoid function
# The function takes:
# - beta: a scalar or array that scales h
# - h: an array of values to apply sigmoid to

beta = 0.5
h = np.array([1.0, 2.0, 3.0, -1.0])

ARGS = {
    "beta": beta,
    "h": h
}
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 (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
swish Ran
ran; our contract draft was wrong, not the code
WarClans612/deep_learning_final/models.py
pointer only (licence: MIT); the stored copy get_code serves, from fengye-lu/PyTorch-SRGAN, is pointer only (licence: GPL-3.0) · get_code("31117990d4f24fdf")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from fengye-lu/PyTorch-SRGAN, and GPL-3.0, as GitHub detects it on that repository, is not in the permissive set Syntology serves inline. 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.

swish Ran
ran; our contract draft was wrong, not the code
islaterm/you-activated-my-neuron/src/neuron/activation_functions.py
pointer only (licence: NONE) · get_code("919aae7a4c5d1c1a")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, islaterm/you-activated-my-neuron 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.

batch_norm_relu Not yet run
unverified (harvested, no recorded run)
Neoanarika/Searching-for-activation-functions/src/childnetwork.py
code served (permissive licence) · get_code("aad048324892b5d3")
How it ran

no invocation recorded.

conv2d_fixed_padding Not yet run
unverified (harvested, no recorded run)
Neoanarika/Searching-for-activation-functions/src/childnetwork.py
code served (permissive licence) · get_code("c756a6986ed2d1b8")
How it ran

no invocation recorded.

create_placeholders Not yet run
unverified (harvested, no recorded run)
WessZumino/expectation_propagation/TF/esp_tf_utils.py
code served (permissive licence) · get_code("7bbbdebd2c04644f")
How it ran

no invocation recorded.

fetch_public_key Not yet run
unverified (harvested, no recorded run)
WarClans612/deep_learning_final/travis_pypi_setup.py
code served (permissive licence) · get_code("4fcc838ec7ac0111")
How it ran

no invocation recorded.

initialize_parameters Not yet run
unverified (harvested, no recorded run)
WessZumino/expectation_propagation/TF/esp_tf_utils.py
code served (permissive licence) · get_code("e1fff8e843c51c8e")
How it ran

no invocation recorded.

one_hot_econding Not yet run
unverified (harvested, no recorded run)
WessZumino/expectation_propagation/TF/esp_tf_utils.py
code served (permissive licence) · get_code("f5bbffdf347b2247")
How it ran

no invocation recorded.

relu_init Not yet run
unverified (harvested, no recorded run)
Neoanarika/Searching-for-activation-functions/src/utils.py
code served (permissive licence) · get_code("c58c0528e0f0b05f")
How it ran

no invocation recorded.

Also present in other papers' code

Of 20 samples lifted from this paper's linked repositories, 20 carry a function identity and 0 do not. Of the 20 distinct functions with an identity, 4 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 3 later) and 16 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
conv2d_fixed_padding 6 other papers
1 dated earlier, 5 later
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
1812.09803 (2018-12)
Guessing Smart: Biased Sampling for Efficient Black-Box Adversarial Attacks
1905.05393 (2019-05)
Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules
and 3 more not named here
fixed_padding 6 other papers
1 dated earlier, 5 later
1512.03385 (2015-12)
Deep Residual Learning for Image Recognition
1812.09803 (2018-12)
Guessing Smart: Biased Sampling for Efficient Black-Box Adversarial Attacks
1905.05393 (2019-05)
Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules
and 3 more not named here
fetch_public_key 1 other paper
1 later
1905.09916 (2019-05)
Generative Grading: Near Human-level Accuracy for Automated Feedback on Richly Structured Problems
swish
brick of this paper's assembly: entry
1 other paper
1 dated earlier
1609.04802 (2016-09)
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

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 choice of activation functions in deep networks has a significant effect on the training dynamics and task performance. Currently, the most successful and widely-used activation function is the Rectified Linear Unit (ReLU). Although various hand-designed alternatives to ReLU have been proposed, none have managed to replace it due to inconsistent gains. In this work, we propose to leverage automatic search techniques to discover new activation functions. Using a combination of exhaustive and reinforcement learning-based search, we discover multiple novel activation functions. We verify the effectiveness of the searches by conducting an empirical evaluation with the best discovered activation function. Our experiments show that the best discovered activation function, f (x) = x • sigmoid(βx), which we name Swish, tends to work better than ReLU on deeper models across a number of challenging datasets. For example, simply replacing ReLUs with Swish units improves top-1 classification accuracy on Im-ageNet by 0.9% for Mobile NASNet-A and 0.6% for Inception-ResNet-v2. The simplicity of Swish and its similarity to ReLU make it easy for practitioners to replace ReLUs with Swish units in any neural network. * Work done as a member of the Google Brain Residency program (g.co/brainresidency).

For agents

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

get_harvested_code_for_paper("1710.05941")
get_code_for_paper("1710.05941")
have("1710.05941")

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

curl https://syntology.ai/api/ran/1710.05941.json

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

[![Syntology run record](https://syntology.ai/api/ran/1710.05941.svg)](https://syntology.ai/paper/1710.05941)

Connect an agent — have() is free.