SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1412.6980 · ICLR · 2014

Adam: A Method for Stochastic Optimization

Diederik Kingma, Jimmy Lei Ba

arXiv · PDF · Open in the Atlas

Code that ran

27 of 37 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (7 of them constructed an object rather than computing a result; 4 honoured the contract we drafted, 2 violated it and 14 have no contract verdict; 7 failures of our instrument; 10 no recorded run), measured 2026-11-21; 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/1412.6980.json · badge: /api/ran/1412.6980.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
KhalilJDamouni/RMSGD_Pretest community (classifier label)
classifier basis not recorded on the link
7 ran · 4 no recorded run
nnaisense/pgpelib community
not in the paper's declared repositories
4 ran · 0 no recorded run
mahdihosseini/AdaS community (classifier label)
classifier basis not recorded on the link
3 ran · 0 no recorded run
rokkian/adam_implementation community
not in the paper's declared repositories
3 ran · 0 no recorded run
navdeepkumar12/Tic community
not in the paper's declared repositories
2 ran · 0 no recorded run
SamuelMarks/doctrans community (classifier label)
classifier basis not recorded on the link
2 ran · 0 no recorded run
vpj/jax_transformer community
not in the paper's declared repositories
1 ran · 1 no recorded run
nyatadecocoa/Machine_Learning_Midterm_Assignment_2019 community
not in the paper's declared repositories
1 ran · 1 no recorded run
MaximeVandegar/Papers-in-100-Lines-of-Code community
not in the paper's declared repositories
1 ran · 0 no recorded run
VAZaytsev/QITE community
Papers with Code archive
1 ran · 0 no recorded run
ylytkin/backpropagation community
not in the paper's declared repositories
1 ran · 0 no recorded run
Sj-Amani/Practical_Behavioral_Cloning community (classifier label)
classifier basis not recorded on the link
1 ran · 0 no recorded run
tomkimpson/ml4l community (classifier label)
classifier basis not recorded on the link
0 ran · 3 no recorded run
labmlai/annotated_deep_learning_paper_implementations community
not in the paper's declared repositories
0 ran · 1 no recorded run
FunctionStatusWhere it lives
Adam Ran
ran on a synthesized input · constructed an object; no result was computed
nnaisense/pgpelib/pgpelib/optimizers.py
code served (permissive licence) · get_code("7f2c925da971c0b5")
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 an instance of its own class, Adam: the run constructed the object and did not call it, so no result was computed. 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

ARGS = {
    "solution_length": 10,
    "dtype": "float32",
    "stepsize": 0.001,
    "beta1": 0.9,
    "beta2": 0.999,
    "epsilon": 1e-8,
}
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 Adam — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Adam 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
Adam Ran
ran on a synthesized input · constructed an object; no result was computed
MaximeVandegar/Papers-in-100-Lines-of-Code/Adam_a_Method_For_Stochastic_Optimization/adam.py
code served (permissive licence) · get_code("cbb1359e9a11e474")
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 an instance of its own class, Adam: the run constructed the object and did not call it, so no result was computed. 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 torch
import torch.nn as nn

torch.manual_seed(0)

# Build a minimal model that has parameters
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear1 = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        return self.linear2(self.linear1(x))

model = SimpleModel()

# Initialize gradients for all parameters
for p in model.parameters():
    p.grad = torch.randn_like(p)

ARGS = {
    "model": model,
    "alpha": 0.001,
    "beta1": 0.9,
    "beta2": 0.999,
    "epsilon": 1e-08
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Adam — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Adam 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
AdamState Ran
ran on a synthesized input · constructed an object; no result was computed
vpj/jax_transformer/transformer.py
pointer only (licence: NONE) · get_code("9dfa8e2f09245ff2")
How it ran

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

Adam_cls Ran
ran on a synthesized input · constructed an object; no result was computed
VAZaytsev/QITE/optimizers.py
pointer only (licence: NONE) · get_code("6f648e6ff626dfd5")
How it ran

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

EVBMF Ran
ran; our contract draft was wrong, not the code
KhalilJDamouni/RMSGD_Pretest/src/adas/VBMF.py
code served (permissive licence) · get_code("8afdc607e63b281b")
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

torch.manual_seed(0)

# Create a small test matrix Y for VBMF
# Shape (L, M) where L <= M
L = 3
M = 5
Y = torch.randn(L, M)

# Construct ARGS dictionary with all parameters for EVBMF
ARGS = {
    'Y': Y,
    'sigma2': None,
    'H': None,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, 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
EVBMF Ran
ran; our contract draft was wrong, not the code
mahdihosseini/AdaS/src/rmsgd/matrix_factorization.py
code served (permissive licence) · get_code("35275a9e3c1edf4f")
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
torch.manual_seed(0)

# Create a small test matrix for Y
# Y should be (L, M) where L <= M
# The function uses torch.svd on Y
L, M = 3, 5
Y = torch.randn(L, M)

# sigma2: noise variance, can be None (will be estimated) or a float
sigma2 = None

# H: maximum rank, can be None (defaults to L) or an int
H = None

ARGS = {
    'Y': Y,
    'sigma2': sigma2,
    'H': H
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, 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
EVBsigma2 Ran
ran, violated the contract we drafted
mahdihosseini/AdaS/src/rmsgd/matrix_factorization.py
code served (permissive licence) · get_code("e2f77eaf75568aa2")
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, 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 numpy as np

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

# Define parameters for EVBsigma2
sigma2 = 1.5  # variance parameter, scalar float
L = 10  # number of latent dimensions
M = 20  # total number of features
s = np.array([0.5, 1.2, 2.1, 0.8, 1.5, 3.2, 0.3, 1.1])  # singular values, 1D array
residual = 5.0  # scalar residual value
xubar = 1.5  # threshold parameter for splitting x values

ARGS = {
    "sigma2": sigma2,
    "L": L,
    "M": M,
    "s": s,
    "residual": residual,
    "xubar": xubar
}
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, 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
NeuralNetwork Ran
ran on a synthesized input · constructed an object; no result was computed
ylytkin/backpropagation/neural_network/neural_network.py
pointer only (licence: NONE) · get_code("7ce4d43450d6e037")
How it ran

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

Optimizer Ran
ran on a synthesized input · constructed an object; no result was computed
nnaisense/pgpelib/pgpelib/optimizers.py
code served (permissive licence) · get_code("2601595a91035c3b")
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 an instance of its own class, Optimizer: the run constructed the object and did not call it, so no result was computed. 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

ARGS = {
    'solution_length': 10,
    'dtype': np.float32,
}
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 Optimizer — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Optimizer 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
VBMF Ran
ran; our contract draft was wrong, not the code
KhalilJDamouni/RMSGD_Pretest/src/adas/VBMF.py
code served (permissive licence) · get_code("0f7078f4c86b344a")
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. 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

# Create a small test matrix Y with shape (L, M) where L <= M
# Using distinct dimensions: 3x5
np.random.seed(0)
Y = np.random.randn(3, 5)

# cacb: product of prior variances - a positive scalar
cacb = 1.0

# sigma2: noise variance - optional, can be None to estimate
sigma2 = None

# H: maximum rank - optional, can be None to default to L
H = None

ARGS = {
    'Y': Y,
    'cacb': cacb,
    'sigma2': sigma2,
    'H': H
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, scipy — 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
VBMF Ran
ran on a synthesized input
KhalilJDamouni/RMSGD_Pretest/src/adas/matrix_factorization.py
code served (permissive licence) · get_code("8f58597b290faee4")
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

# Create a small test matrix for factorization
# Y should be (L, M) where L <= M
np.random.seed(0)
L, M = 3, 5
# Create a low-rank matrix for realistic SVD
U_true = np.random.randn(L, 2)
V_true = np.random.randn(M, 2)
Y = U_true @ V_true.T + 0.1 * np.random.randn(L, M)

# Prior variance product
cacb = 1.0

# sigma2 and H can be None (they will be estimated/set automatically)
sigma2 = None
H = None

ARGS = {
    "Y": Y,
    "cacb": cacb,
    "sigma2": sigma2,
    "H": H,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — 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
VBsigma2 Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
KhalilJDamouni/RMSGD_Pretest/src/adas/VBMF.py
code served (permissive licence) · get_code("9e8ce2d162e22cdd")
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

# Analyze the function to understand parameter requirements:
# - sigma2: scalar float (noise variance)
# - L: int (appears in calculations)
# - M: int (appears in calculations)
# - cacb: scalar float (appears in divisions and multiplications)
# - s: 1D numpy array of singular values (sorted in descending order)
# - residual: scalar float (residual sum of squares)

# Create small test data
np.random.seed(0)

# s should be a 1D array of singular values, typically sorted descending
s = np.array([5.0, 3.0, 1.5, 0.5])

# Scalar parameters
sigma2 = 1.0
L = 3
M = 4
cacb = 2.0
residual = 2.5

ARGS = {
    "sigma2": sigma2,
    "L": L,
    "M": M,
    "cacb": cacb,
    "s": s,
    "residual": residual
}
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
adam Ran
ran on a synthesized input
navdeepkumar12/Tic/nn.py
pointer only (licence: NONE) · get_code("d617942446009de2")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, navdeepkumar12/Tic 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.

adam Ran
ran; our contract draft was wrong, not the code
rokkian/adam_implementation/adam_implementation.py
code served (permissive licence) · get_code("da1304a720259f4d")
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 list, ndim 2, 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.

from math import sqrt
from numpy import asarray, arange, meshgrid
from numpy.random import rand, seed

def derivative(x, y):
    """Gradient of the covnex Objective function: z'=2x + 2y"""
    return asarray([x * 2.0, y * 2.0])

def objective(x, y):
    """funzione obiettivo z = x**2 + y**2 """
    return x ** 2.0 + y ** 2.0

def adam(objective, derivative, bounds, n_iter, alpha, beta1, beta2, eps=1e-08):
    """Adam alg"""
    solutions = list()
    x = bounds[:, 0] + rand(len(bounds)) * (bounds[:, 1] - bounds[:, 0])
    score = objective(x[0], x[1])
    m = [0.0 for _ in range(bounds.shape[0])]
    v = [0.0 for _ in range(bounds.shape[0])]
    for t in range(n_iter):
        g = derivative(x[0], x[1])
        for i in range(bounds.shape[0]):
            m[i] = beta1 * m[i] + (1.0 - beta1) * g[i]
            v[i] = beta2 * v[i] + (1.0 - beta2) * g[i] ** 2
            mhat = m[i] / (1 - beta1 ** (t + 1))
            vhat = v[i] / (1 - beta2 ** (t + 1))
            x[i] = x[i] - alpha * mhat / (sqrt(vhat) + eps)
        score = objective(x[0], x[1])
        solutions.append(x.copy())
        print(f'>{t} f({str(x)}) = {score:.5f}')
    return solutions

import numpy as np

seed(42)

bounds = np.asarray([[-5.0, 5.0], [-5.0, 5.0]])
n_iter = 3
alpha = 0.001
beta1 = 0.9
beta2 = 0.999

ARGS = {
    'objective': objective,
    'derivative': derivative,
    'bounds': bounds,
    'n_iter': n_iter,
    'alpha': alpha,
    'beta1': beta1,
    'beta2': beta2,
    'eps': 1e-08
}
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 list, ndim 2, dtype family float — 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
alexnet Ran
ran; our contract draft was wrong, not the code
KhalilJDamouni/RMSGD_Pretest/src/adas/models/alexnet.py
code served (permissive licence) · get_code("cd7d9260519b0c88")
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 AlexNet. 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 torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'pretrained': False,
    'progress': True,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type AlexNet — a AlexNet output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a AlexNet 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
clear_annotation Ran
ran on a synthesized input
SamuelMarks/doctrans/cdd/compound/doctrans_utils.py
code served (permissive licence) · get_code("d05c0d4ad3893add")
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 AnnAssign. 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 ast

# Create an AST node with annotation and type_comment attributes
node = ast.AnnAssign(
    target=ast.Name(id='x', ctx=ast.Store()),
    annotation=ast.Name(id='int', ctx=ast.Load()),
    value=None,
    simple=1
)
node.type_comment = "int"

ARGS = {
    'node': node
}
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 AnnAssign — a AnnAssign output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a AnnAssign 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
derivative Ran
ran, honoured the contract we drafted
rokkian/adam_implementation/adam_implementation.py
code served (permissive licence) · get_code("4618e61672a36671")
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 ndarray, ndim 1, dtype family float, shape (2,) 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

# The function takes two scalar parameters x and y
# and computes their gradients in a simple convex objective function
# The function multiplies each parameter by 2.0

x_val = 3.5
y_val = 2.1

ARGS = {
    "x": x_val,
    "y": y_val,
}
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 (2,) — shape (2,) 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
has_type_annotations Ran
ran on a synthesized input
SamuelMarks/doctrans/cdd/compound/doctrans_utils.py
code served (permissive licence) · get_code("b8d84bad5cfe2f9c")
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 ast
from ast import AST, AnnAssign, Assign, AsyncFunctionDef, ClassDef, FunctionDef, NodeTransformer, arg, get_docstring, walk

# Create a simple AST node with type annotations
# Option 1: Create a FunctionDef with return annotation
func_node = FunctionDef(
    name="example_func",
    args=ast.arguments(
        posonlyargs=[],
        args=[arg(arg="x", annotation=ast.Name(id="int"))],
        kwonlyargs=[],
        kw_defaults=[],
        defaults=[]
    ),
    body=[ast.Pass()],
    decorator_list=[],
    returns=ast.Name(id="str")
)

ARGS = {
    "node": func_node
}
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
moving_average Ran
ran on a synthesized input
Sj-Amani/Practical_Behavioral_Cloning/left_right_driving_data_preprocessing.py
code served (permissive licence) · get_code("755b233b6180c3d0")
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 (8,) 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 numpy array for testing
test_array = np.array([1.0, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, 9.0, 10.0])

# Define window size for moving average
window_size = 3

ARGS = {
    "a": test_array,
    "n": window_size
}
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 (8,) — shape (8,) 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
objective Ran
ran, honoured the contract we drafted
rokkian/adam_implementation/adam_implementation.py
code served (permissive licence) · get_code("f6c61d4de559ffe9")
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 float, 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 torch
import numpy as np

torch.manual_seed(0)

# The function takes two parameters: x and y
# Both are used in arithmetic operations (squaring and addition)
# They should be numeric types - can be float, int, or tensors

x = 3.0
y = 4.0

ARGS = {
    "x": x,
    "y": y,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
optimizer Ran
ran on a synthesized input · constructed an object; no result was computed
navdeepkumar12/Tic/nn.py
pointer only (licence: NONE) · get_code("77da68eb2be45988")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, navdeepkumar12/Tic 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.

phi0 Ran
ran, violated the contract we drafted
mahdihosseini/AdaS/src/rmsgd/matrix_factorization.py
code served (permissive licence) · get_code("2760d409b48cf23f")
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
import torch

torch.manual_seed(0)

# phi0 takes a single parameter x
# The function computes x - np.log(x)
# x must support subtraction and be compatible with np.log()
# This works with numpy arrays or scalar values

# Create a small numpy array as input
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, through the batch API
Imports
numpy, torch — 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
positive_float Ran
ran, honoured the contract we drafted
nnaisense/pgpelib/pgpelib/optimizers.py
code served (permissive licence) · get_code("2425339890ade98d")
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 float, 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 torch
import numpy as np

# positive_float expects a single parameter 'x' that can be converted to float
# and must be positive (> 0)

ARGS = {
    'x': 3.14
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
positive_int Ran
ran, honoured the contract we drafted
nnaisense/pgpelib/pgpelib/optimizers.py
code served (permissive licence) · get_code("3c047cbc5f859f9b")
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 int, ndim 0, dtype family int, 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 torch

torch.manual_seed(0)

# The function positive_int takes a parameter x and converts it to an int,
# then checks if it's positive. It can accept integers, floats, strings, etc.
# that are convertible to int.

ARGS = {
    'x': 5
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, 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
safe_float_cast Ran
ran on a synthesized input
KhalilJDamouni/RMSGD_Pretest/src/adas/utils.py
code served (permissive licence) · get_code("9a5925e41635d417")
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 float, 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 torch

ARGS = {
    "str_number": "3.14"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
smart_string_to_float Ran
ran on a synthesized input
KhalilJDamouni/RMSGD_Pretest/src/adas/utils.py
code served (permissive licence) · get_code("0a32bb4b38232251")
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 float, 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 torch

torch.manual_seed(0)

ARGS = {
    'string': '3.14',
    'e': 'could not convert string to float'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type float, 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
st_ops Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
nyatadecocoa/Machine_Learning_Midterm_Assignment_2019/Problem7/adam.py
pointer only (licence: NONE) · get_code("62c36ccd5072fcd8")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, nyatadecocoa/Machine_Learning_Midterm_Assignment_2019 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.

Adam Not yet run
unverified (harvested, no recorded run)
vpj/jax_transformer/transformer.py
pointer only (licence: NONE) · get_code("c48006199ed3e050")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, vpj/jax_transformer 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. No invocation recorded.

Adam Not yet run
unverified (harvested, no recorded run)
labmlai/annotated_deep_learning_paper_implementations/labml_nn/optimizers/adam.py
code served (permissive licence) · get_code("10c32bb6eaf9662a")
How it ran

no invocation recorded.

get_data Not yet run
unverified (harvested, no recorded run)
KhalilJDamouni/RMSGD_Pretest/src/adas/data.py
code served (permissive licence) · get_code("ab7b976a4eb5340d")
How it ran

no invocation recorded.

load_meta_file Not yet run
unverified (harvested, no recorded run)
KhalilJDamouni/RMSGD_Pretest/src/adas/datasets.py
code served (permissive licence) · get_code("5e1f354096a8c5bc")
How it ran

no invocation recorded.

optimize Not yet run
unverified (harvested, no recorded run)
nyatadecocoa/Machine_Learning_Midterm_Assignment_2019/Problem7/adam.py
pointer only (licence: NONE) · get_code("c5c5da8da7132910")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, nyatadecocoa/Machine_Learning_Midterm_Assignment_2019 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. No invocation recorded.

parse_devkit_archive Not yet run
unverified (harvested, no recorded run)
KhalilJDamouni/RMSGD_Pretest/src/adas/datasets.py
code served (permissive licence) · get_code("6257f483693ebe01")
How it ran

no invocation recorded.

process_MODIS_file Not yet run
unverified (harvested, no recorded run)
tomkimpson/ml4l/legacy/legacy_scripts/join_MODIS_with_ERA_V2.py
code served (permissive licence) · get_code("1a88d05a828e9b9c")
How it ran

no invocation recorded.

pstats_to_dict Not yet run
unverified (harvested, no recorded run)
KhalilJDamouni/RMSGD_Pretest/src/adas/utils.py
code served (permissive licence) · get_code("fd7ae3f51b2d34cd")
How it ran

no invocation recorded.

select_correct_MODIS_file Not yet run
unverified (harvested, no recorded run)
tomkimpson/ml4l/legacy/legacy_scripts/join_MODIS_with_ERA_V2.py
code served (permissive licence) · get_code("a3a5ccc0ea8274ee")
How it ran

no invocation recorded.

train_NN Not yet run
unverified (harvested, no recorded run)
tomkimpson/ml4l/legacy/legacy_scripts/feature_importance.py
code served (permissive licence) · get_code("540ea376e18a8af6")
How it ran

no invocation recorded.

Also present in other papers' code

Of 37 samples lifted from this paper's linked repositories, 37 carry a function identity and 0 do not. Of the 37 distinct functions with an identity, 9 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 9 later) and 28 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
alexnet 4 other papers
4 later
1910.00033 (2019-10)
Hidden Trigger Backdoor Attacks
2010.09080 (2020-10)
Poisoned classifiers are not only backdoored, they are fundamentally broken
2305.19101 (2023-05)
Which Models have Perceptually-Aligned Gradients? An Explanation via Off-Manifold Robustness
and 1 more not named here
VBMF 2 other papers
2 later
2108.06822 (2021-08)
CONet: Channel Optimization for Convolutional Neural Networks
2111.14056 (2021-11)
Towards Robust and Automatic Hyper-Parameter Tunning
VBsigma2 2 other papers
2 later
2108.06822 (2021-08)
CONet: Channel Optimization for Convolutional Neural Networks
2111.14056 (2021-11)
Towards Robust and Automatic Hyper-Parameter Tunning
EVBMF 1 other paper
1 later
2110.12259 (2021-10)
In Search of Probeable Generalization Measures
EVBMF 1 other paper
1 later
2108.06822 (2021-08)
CONet: Channel Optimization for Convolutional Neural Networks
EVBsigma2 1 other paper
1 later
2110.12259 (2021-10)
In Search of Probeable Generalization Measures
load_meta_file 1 other paper
1 later
2108.06822 (2021-08)
CONet: Channel Optimization for Convolutional Neural Networks
parse_devkit_archive 1 other paper
1 later
2108.06822 (2021-08)
CONet: Channel Optimization for Convolutional Neural Networks
phi0 1 other paper
1 later
2110.12259 (2021-10)
In Search of Probeable Generalization Measures

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 2026-11-21.

Repositories linked to this paper

More are linked than shown here.

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

Abstract

We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and/or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. Empirical results demonstrate that Adam works well in practice and compares favorably to other stochastic optimization methods. Finally, we discuss AdaMax, a variant of Adam based on the infinity norm. * Equal contribution. Author ordering determined by coin flip over a Google Hangout.

For agents

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

get_harvested_code_for_paper("1412.6980")
get_code_for_paper("1412.6980")
have("1412.6980")

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

curl https://syntology.ai/api/ran/1412.6980.json

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

[![Syntology run record](https://syntology.ai/api/ran/1412.6980.svg)](https://syntology.ai/paper/1412.6980)

Connect an agent — have() is free.