SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 1106.5730 · 2011

HOGWILD!: A Lock-Free Approach to Parallelizing Stochastic Gradient Descent

arXiv · PDF · Open in the Atlas

Code that ran

16 of 24 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, 0 violated it and 16 have no contract verdict; 0 failures of our instrument; 8 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/1106.5730.json · badge: /api/ran/1106.5730.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
lifeomic/sparkflow community (classifier label)
classifier basis not recorded on the link
6 ran · 4 no recorded run
maxpumperla/elephas community (classifier label)
classifier basis not recorded on the link
6 ran · 0 no recorded run
dmmiller612/sparktorch community (classifier label)
classifier basis not recorded on the link
4 ran · 4 no recorded run
FunctionStatusWhere it lives
add_params Ran
ran on a synthesized input
maxpumperla/elephas/elephas/utils/functional_utils.py
code served (permissive licence) · get_code("5845a9f358c1aebd")
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. 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 two lists of numpy arrays with different shapes for clarity
param_list_left = [
    np.array([1.0, 2.0, 3.0, 4.0]),
    np.array([[1.0, 2.0], [3.0, 4.0]]),
    np.array([[[1.0, 2.0, 3.0]]])
]

param_list_right = [
    np.array([0.5, 0.5, 0.5, 0.5]),
    np.array([[0.1, 0.2], [0.3, 0.4]]),
    np.array([[[0.1, 0.2, 0.3]]])
]

ARGS = {
    'param_list_left': param_list_left,
    'param_list_right': param_list_right
}
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 — a list 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
as_enum Ran
ran on a synthesized input
maxpumperla/elephas/elephas/utils/model_utils.py
code served (permissive licence) · get_code("26266f994d9c5ab4")
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 ModelType. 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
from enum import Enum

class ModelType(Enum):
    CLASSIFICATION = 1
    REGRESSION = 2

def as_enum(d):
    if '__enum__' in d:
        name, member = d['__enum__'].split('.')
        return getattr(ModelType, member)
    else:
        return d

# Test case 1: dict with __enum__ key
d_with_enum = {
    '__enum__': 'ModelType.CLASSIFICATION'
}

ARGS = {
    'd': d_with_enum
}
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 ModelType — a ModelType output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a ModelType 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
build_adam_config Ran
ran on a synthesized input
lifeomic/sparkflow/sparkflow/graph_utils.py
code served (permissive licence) · get_code("f1d06b5ecd11b217")
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 torch
import json

torch.manual_seed(0)

# The function build_adam_config takes 5 parameters, all with default values
# It constructs a config by calling generate_config which returns a JSON string

ARGS = {
    "learning_rate": 0.001,
    "beta1": 0.9,
    "beta2": 0.999,
    "epsilon": 1e-08,
    "use_locking": False
}
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 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
build_rmsprop_config Ran
ran on a synthesized input
lifeomic/sparkflow/sparkflow/graph_utils.py
code served (permissive licence) · get_code("6931025bf1d0e11b")
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 torch
import json

torch.manual_seed(0)

# The function build_rmsprop_config accepts the following parameters:
# - learning_rate: float (default 0.001)
# - decay: float (default 0.9)
# - momentum: float (default 0.0)
# - epsilon: float (default 1e-10)
# - use_locking: bool (default False)
# - centered: bool (default False)

# The function simply passes these to generate_config which returns a JSON string.
# No complex objects are needed - just scalar values.

ARGS = {
    "learning_rate": 0.001,
    "decay": 0.9,
    "momentum": 0.0,
    "epsilon": 1e-10,
    "use_locking": False,
    "centered": False,
}
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 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
convert_json_to_weights Ran
ran on a synthesized input
lifeomic/sparkflow/sparkflow/ml_util.py
code served (permissive licence) · get_code("b665a4bab47422a1")
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. 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 json

# Construct valid arguments for convert_json_to_weights
# The function expects a JSON string containing weights
# that will be parsed and converted to numpy arrays

# Create sample weight data
weights_data = [
    [[1.0, 2.0, 3.0, 4.0],
     [5.0, 6.0, 7.0, 8.0]],
    [[9.0, 10.0, 11.0, 12.0],
     [13.0, 14.0, 15.0, 16.0]],
    [0.5, 0.6, 0.7, 0.8]
]

# Convert to JSON string
json_weights = json.dumps(weights_data)

ARGS = {
    "json_weights": json_weights
}
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 — a list 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
convert_to_serialized_torch Ran
ran on a synthesized input
dmmiller612/sparktorch/sparktorch/inference.py
code served (permissive licence) · get_code("f7af48c85d7f2b6e")
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 torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleNetwork(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
        self.relu = nn.ReLU()
    
    def forward(self, x):
        x = self.linear(x)
        x = self.relu(x)
        return x

network = SimpleNetwork()

ARGS = {
    'network': network
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
dill, torch — 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
convert_weights_to_json Ran
ran on a synthesized input
lifeomic/sparkflow/sparkflow/ml_util.py
code served (permissive licence) · get_code("2fc09059ad3fca01")
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 torch
import numpy as np

torch.manual_seed(0)

# The function expects a list of objects that have a .tolist() method
# This means the objects should be torch tensors or numpy arrays
# Create a list of small tensors with distinct shapes

weights = [
    torch.randn(2, 4),
    torch.randn(3, 5),
    torch.randn(2)
]

ARGS = {
    "weights": weights
}
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 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
from_matrix Ran
ran on a synthesized input
maxpumperla/elephas/elephas/mllib/adapter.py
code served (permissive licence) · get_code("2712866b8caee6bd")
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 (4, 3) 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 mock MLlib Matrix class that has a toArray() method
class MockMLlibMatrix:
    def __init__(self, data):
        self._data = np.array(data)
    
    def toArray(self):
        return self._data

# Construct a small matrix with distinct dimensions (4 rows, 3 columns)
matrix_data = [
    [1.0, 2.0, 3.0],
    [4.0, 5.0, 6.0],
    [7.0, 8.0, 9.0],
    [10.0, 11.0, 12.0]
]

mock_matrix = MockMLlibMatrix(matrix_data)

ARGS = {
    "matrix": mock_matrix
}
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 (4, 3) — shape (4, 3) 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
from_vector Ran
ran on a synthesized input
maxpumperla/elephas/elephas/mllib/adapter.py
code served (permissive licence) · get_code("80ae39b85048fd6f")
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
try:
    from pyspark.mllib.linalg import DenseVector
except ImportError:
    # Fallback: create a minimal mock that has toArray() method
    class DenseVector:
        def __init__(self, values):
            self.values = np.array(values)
        
        def toArray(self):
            return self.values

# Create a small vector with distinct values
vector_data = np.array([1.5, 2.3, 3.7, 4.2])
vector = DenseVector(vector_data)

ARGS = {
    "vector": vector
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pyspark — 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
get_neutral Ran
ran on a synthesized input
maxpumperla/elephas/elephas/utils/functional_utils.py
code served (permissive licence) · get_code("dabcb1f56c2aeec4")
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. 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 list of numpy arrays with different shapes
array1 = np.array([[1, 2, 3], [4, 5, 6]])
array2 = np.array([7, 8, 9, 10])
array3 = np.ones((2, 3, 4))

array_list = [array1, array2, array3]

ARGS = {
    "array_list": array_list
}
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 — a list 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
handle_data Ran
ran on a synthesized input
lifeomic/sparkflow/sparkflow/tensorflow_async.py
code served (permissive licence) · get_code("2f23661733396fbf")
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 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.

import numpy as np

# Create a simple dictionary-like data structure with column access
class SimpleData:
    def __init__(self):
        self.data = {
            'input_column': np.array([1.0, 2.0, 3.0, 4.0]),
            'label_column': np.array([10, 20, 30, 40])
        }
    
    def __getitem__(self, key):
        return self.data[key]

# Create test data
data = SimpleData()

# Define arguments for the function
ARGS = {
    'data': data,
    'inp_col': 'input_column',
    'label_col': 'label_column'
}
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, ndim 2, 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
load_byte_array Ran
ran on a synthesized input
dmmiller612/sparktorch/sparktorch/pipeline_util.py
code served (permissive licence) · get_code("9d11d280b0167a20")
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 sys
import zlib
import dill
import torch
import numpy as np

# Create a simple Python object to serialize
test_obj = {"key": "value", "number": 42, "list": [1, 2, 3]}

# Serialize the object using dill and zlib
serialized = dill.dumps(test_obj)
compressed = zlib.compress(serialized)

# Convert compressed bytes to a comma-separated string of byte values
byte_values = [str(b) for b in compressed]
byte_string = ','.join(byte_values) + ','

ARGS = {
    'stop_words': [byte_string]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
dill, numpy, torch — 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
load_byte_array Ran
ran on a synthesized input
lifeomic/sparkflow/sparkflow/pipeline_util.py
code served (permissive licence) · get_code("3974c66cb4e0844e")
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 dill
import zlib
import sys

# Create a simple Python object to serialize
test_obj = {"key": "value", "number": 42, "list": [1, 2, 3]}

# Serialize it using dill and compress with zlib
serialized = dill.dumps(test_obj)
compressed = zlib.compress(serialized)

# Convert compressed bytes to comma-separated string of byte values
byte_values = ','.join(str(b) for b in compressed)
# Add trailing comma to match the split()[0:-1] logic in the function
stop_words_string = byte_values + ','

ARGS = {
    "stop_words": [stop_words_string]
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
dill — 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
subtract_params Ran
ran on a synthesized input
maxpumperla/elephas/elephas/utils/functional_utils.py
code served (permissive licence) · get_code("2bd77115365b2003")
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. 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 two lists of numpy arrays with different sizes to test the function
param_list_left = [
    np.array([[1.0, 2.0, 3.0, 4.0],
              [5.0, 6.0, 7.0, 8.0]]),
    np.array([0.5, 1.5, 2.5, 3.5]),
    np.array([[[10.0, 20.0], [30.0, 40.0]]])
]

param_list_right = [
    np.array([[0.1, 0.2, 0.3, 0.4],
              [0.5, 0.6, 0.7, 0.8]]),
    np.array([0.1, 0.1, 0.1, 0.1]),
    np.array([[[1.0, 2.0], [3.0, 4.0]]])
]

ARGS = {
    "param_list_left": param_list_left,
    "param_list_right": param_list_right
}
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 — a list 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
torch_decoder Ran
ran on a synthesized input
dmmiller612/sparktorch/sparktorch/util.py
code served (permissive licence) · get_code("e4fd2bc4dbc289b1")
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 SimpleModel. 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
import codecs
import dill

torch.manual_seed(0)

# Create a simple model to serialize
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# Create and serialize a model
model = SimpleModel()

# Serialize the model using dill and base64 encoding
model_bytes = dill.dumps(model)
model_ser = codecs.encode(model_bytes, 'base64').decode()

ARGS = {
    'model_ser': model_ser
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
dill, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type SimpleModel — a SimpleModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleModel 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
torch_encoder Ran
ran on a synthesized input
dmmiller612/sparktorch/sparktorch/util.py
code served (permissive licence) · get_code("899235f37bc817b4")
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 torch
import torch.nn as nn
import codecs
import dill

torch.manual_seed(0)

# The function torch_encoder takes any object and serializes it using dill + base64
# It doesn't actually use the object in any special way, just serializes it
# So we can pass various types of objects

# Let's create a simple torch tensor as the object to encode
test_tensor = torch.randn(2, 4)

ARGS = {
    'obj': test_tensor
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
dill, torch — 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
build_graph Not yet run
unverified (harvested, no recorded run)
lifeomic/sparkflow/sparkflow/graph_utils.py
code served (permissive licence) · get_code("de19052b4807c568")
How it ran

no invocation recorded.

build_optimizer Not yet run
unverified (harvested, no recorded run)
lifeomic/sparkflow/sparkflow/tensorflow_async.py
code served (permissive licence) · get_code("65ad80c13f16808e")
How it ran

no invocation recorded.

get_main Not yet run
unverified (harvested, no recorded run)
dmmiller612/sparktorch/sparktorch/hogwild.py
code served (permissive licence) · get_code("60068090f95c632a")
How it ran

no invocation recorded.

get_server_weights Not yet run
unverified (harvested, no recorded run)
lifeomic/sparkflow/sparkflow/HogwildSparkModel.py
code served (permissive licence) · get_code("a17f3090c2ef6a4c")
How it ran

no invocation recorded.

get_state_dict Not yet run
unverified (harvested, no recorded run)
dmmiller612/sparktorch/sparktorch/hogwild.py
code served (permissive licence) · get_code("1146a0a7466202aa")
How it ran

no invocation recorded.

handle_features Not yet run
unverified (harvested, no recorded run)
dmmiller612/sparktorch/sparktorch/util.py
code served (permissive licence) · get_code("1bdf839eca414ecc")
How it ran

no invocation recorded.

put_early_stop Not yet run
unverified (harvested, no recorded run)
dmmiller612/sparktorch/sparktorch/hogwild.py
code served (permissive licence) · get_code("4afd4babdb245f0e")
How it ran

no invocation recorded.

tensorflow_get_weights Not yet run
unverified (harvested, no recorded run)
lifeomic/sparkflow/sparkflow/ml_util.py
code served (permissive licence) · get_code("39f1bc951fa89a77")
How it ran

no invocation recorded.

Also present in other papers' code

Of 24 samples lifted from this paper's linked repositories, 24 carry a function identity and 0 do not. Of the 23 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 23 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.

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

Stochastic Gradient Descent (SGD) is a popular algorithm that can achieve state-of-the-art performance on a variety of machine learning tasks. Several researchers have recently proposed schemes to parallelize SGD, but all require performance-destroying memory locking and synchronization. This work aims to show using novel theoretical analysis, algorithms, and implementation that SGD can be implemented without any locking. We present an update scheme called HOGWILD! which allows processors access to shared memory with the possibility of overwriting each other's work. We show that when the associated optimization problem is sparse, meaning most gradient updates only modify small parts of the decision variable, then HOGWILD! achieves a nearly optimal rate of convergence. We demonstrate experimentally that HOGWILD! outperforms alternative schemes that use locking by an order of magnitude.

For agents

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

get_harvested_code_for_paper("1106.5730")
get_code_for_paper("1106.5730")
have("1106.5730")

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

curl https://syntology.ai/api/ran/1106.5730.json

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

[![Syntology run record](https://syntology.ai/api/ran/1106.5730.svg)](https://syntology.ai/paper/1106.5730)

Connect an agent — have() is free.