Paper · 1906.02773 · 2019
One ticket to win them all: generalizing lottery ticket initializations across datasets and optimizers
arXiv · PDF · Open in the Atlas
Code that ran
4 of 4 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, 1 violated it and 0 have no contract verdict; 3 failures of our instrument; 0 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.
- 1 ran with no instrument failure
0 honoured the contract we drafted, 1 violated it, 0 have no contract verdict
- 3 ran where Syntology's instrument failed
failures of Syntology's instrument, not of the code: 1 where our contract draft was wrong, 2 where the contract check could not run
- 0 have no recorded run
- 0 of those that ran constructed an object rather than computing a result
an executed sample whose run returned an instance of its own class (fixture_out_type equals the entry name): the run built an object and did not compute a result
- 0 are pointer only: the body is not served under this paper's link
"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/1906.02773.json · badge: /api/ran/1906.02773.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.
| Function | Status | Where it lives |
|---|
| get_20_percent |
Ran ran, violated the contract we drafted |
varungohil/Generalizing-Lottery-Tickets/src/iterative_pruning.py code served (permissive licence) · get_code("c48f350f6d321cd0") How it ranRan in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) 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 get_20_percent takes a single parameter 'total' which should be a number
# It simply multiplies the input by 0.2 and returns the result
# So we need to provide a numeric value
ARGS = {
"total": 100
}
- Setup block drafted by
- a language model, Claude Haiku 4.5, prompt python_fixture_v1
- 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
|
| modelselect |
Ran ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT) |
ismail31416/colt/main-full.py code served (permissive licence) · get_code("4efe71ec2340b424") How it ranRan 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 MockModule. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
from argparse import Namespace
torch.manual_seed(0)
# Create a minimal args object with required attributes
args = Namespace(
dataset='mnist',
arch_type='fc1'
)
# Create mock architecture modules to avoid import errors
class MockModule(nn.Module):
def __init__(self, num_classes=None):
super().__init__()
self.linear = nn.Linear(10, num_classes or 10)
def forward(self, x):
return self.linear(x)
class fc1:
@staticmethod
def fc1(num_classes=None):
return MockModule(num_classes)
class LeNet5:
@staticmethod
def LeNet5(num_classes=None):
return MockModule(num_classes)
class AlexNet:
@staticmethod
def AlexNet(num_classes=None):
return MockModule(num_classes)
class resnet:
@staticmethod
def resnet18(num_classes=None):
return MockModule(num_classes)
class densenet:
@staticmethod
def densenet121(num_classes=None):
return MockModule(num_classes)
class conv3:
@staticmethod
def conv3(num_classes=None):
return MockModule(num_classes)
class mobilenetv2:
@staticmethod
def mobilenetv2(num_classes=None):
return MockModule(num_classes)
class shufflenetv2:
@staticmethod
def shufflenetv2(num_classes=None):
return MockModule(num_classes)
# Mock the archs module structure
import sys
sys.modules['archs'] = type(sys)('archs')
sys.modules['archs.mnist'] = type(sys)('archs.mnist')
sys.modules['archs.cifar10'] = type(sys)('archs.cifar10')
sys.modules['archs.fashionmnist'] = type(sys)('archs.fashionmnist')
sys.modules['archs.cifar100'] = type(sys)('archs.cifar100')
sys.modules['archs.imagenet'] = type(sys)('archs.imagenet')
sys.modules['archs.tinyimagenet'] = type(sys)('archs.tinyimagenet')
for arch_module in ['archs.mnist', 'archs.cifar10', 'archs.fashionmnist', 'archs.cifar100', 'archs.imagenet', 'archs.tinyimagenet']:
sys.modules[arch_module].fc1 = fc1
sys.modules[arch_module].LeNet5 = LeNet5
sys.modules[arch_module].AlexNet = AlexNet
sys.modules[arch_module].resnet = resnet
sys.modules[arch_module].densenet = densenet
sys.modules[arch_module].conv3 = conv3
sys.modules[arch_module].mobilenetv2 = mobilenetv2
sys.modules[arch_module].shufflenetv2 = shufflenetv2
ARGS = {
'args': args,
'output_class': 10
}
- Setup block drafted by
- a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
- Imports
- archs, torch — import names as written, not package names to install; a name the repository itself provides is not a package
- Returned
- type MockModule — a MockModule output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MockModule 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
|
| modelselect |
Ran ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT) |
ismail31416/colt/transfer.py code served (permissive licence) · get_code("895e488cbc2f1d73") How it ranRan in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned MockModule. The image it ran in was not recorded. The code mentions cuda, and the harness aliases cuda to the CPU for every sample; whether this run needed that is 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
from unittest.mock import MagicMock
torch.manual_seed(0)
# Create a minimal args object with required attributes
class Args:
def __init__(self):
self.dataset1 = 'mnist'
self.arch_type = 'fc1'
# Create mock modules for the architecture imports
# These will be imported by modelselect, so we need to mock them
class MockModule(nn.Module):
def __init__(self, num_classes=None):
super().__init__()
self.linear = nn.Linear(10, num_classes if num_classes else 10)
def forward(self, x):
return self.linear(x)
# Create mock architecture modules
mock_fc1 = MagicMock()
mock_fc1.fc1 = MockModule
mock_lenet5 = MagicMock()
mock_lenet5.LeNet5 = MockModule
mock_alexnet = MagicMock()
mock_alexnet.AlexNet = MockModule
mock_resnet = MagicMock()
mock_resnet.resnet18 = MockModule
mock_densenet = MagicMock()
mock_densenet.densenet121 = MockModule
mock_conv3 = MagicMock()
mock_conv3.conv3 = MockModule
mock_mobilenetv2 = MagicMock()
mock_mobilenetv2.mobilenetv2 = MockModule
mock_shufflenetv2 = MagicMock()
mock_shufflenetv2.shufflenetv2 = MockModule
# Patch the imports in sys.modules to avoid actual file imports
import sys
sys.modules['archs'] = MagicMock()
sys.modules['archs.mnist'] = MagicMock(
fc1=mock_fc1,
LeNet5=mock_lenet5,
AlexNet=mock_alexnet,
resnet=mock_resnet,
densenet=mock_densenet,
conv3=mock_conv3,
mobilenetv2=mock_mobilenetv2,
shufflenetv2=mock_shufflenetv2
)
sys.modules['archs.cifar10'] = sys.modules['archs.mnist']
sys.modules['archs.cifar100'] = sys.modules['archs.mnist']
sys.modules['archs.imagenet'] = sys.modules['archs.mnist']
sys.modules['archs.tinyimagenet'] = sys.modules['archs.mnist']
ARGS = {
'args': Args(),
'output_class': 10
}
- Setup block drafted by
- a language model, Claude Haiku 4.5, prompt python_fixture_v1
- Imports
- archs, torch — import names as written, not package names to install; a name the repository itself provides is not a package
- Returned
- type MockModule — a MockModule output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MockModule 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
|
| permute_masks |
Ran ran; our contract draft was wrong, not the code |
varungohil/Generalizing-Lottery-Tickets/src/iterative_pruning.py code served (permissive licence) · get_code("e33fdf660fdbaec4") How it ranRan in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) 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 torch
import numpy as np
torch.manual_seed(0)
np.random.seed(0)
# Create sample masks as tensors with different shapes
mask1 = torch.randint(0, 2, (4, 3)) # 12 elements
mask2 = torch.randint(0, 2, (2, 5)) # 10 elements
mask3 = torch.randint(0, 2, (3,)) # 3 elements
old_masks = [mask1, mask2, mask3]
ARGS = {
"old_masks": old_masks
}
- Setup block drafted by
- a language model, Claude Haiku 4.5, prompt python_fixture_v1
- Imports
- numpy, torch — 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
|
Also present in other papers' code
Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).
Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.
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 success of lottery ticket initializations (Frankle and Carbin, 2019) suggests that small, sparsified networks can be trained so long as the network is initialized appropriately. Unfortunately, finding these "winning ticket" initializations is computationally expensive. One potential solution is to reuse the same winning tickets across a variety of datasets and optimizers. However, the generality of winning ticket initializations remains unclear. Here, we attempt to answer this question by generating winning tickets for one training configuration (optimizer and dataset) and evaluating their performance on another configuration. Perhaps surprisingly, we found that, within the natural images domain, winning ticket initializations generalized across a variety of datasets, including Fashion MNIST, SVHN, CIFAR-10/100, ImageNet, and Places365, often achieving performance close to that of winning tickets generated on the same dataset. Moreover, winning tickets generated using larger datasets consistently transferred better than those generated using smaller datasets. We also found that winning ticket initializations generalize across optimizers with high performance. These results suggest that winning ticket initializations generated by sufficiently large datasets contain inductive biases generic to neural networks more broadly which improve training across many settings and provide hope for the development of better initialization methods.
For agents
The same record, over MCP at https://syntology.ai/mcp:
get_harvested_code_for_paper("1906.02773")
get_code_for_paper("1906.02773")
have("1906.02773")
The run record, dated, one paper per request, free:
curl https://syntology.ai/api/ran/1906.02773.json
A badge for a README (the split and the date, never a ratio):
[](https://syntology.ai/paper/1906.02773)
Connect an agent — have() is free.