SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2010.11560 · 2020

Deep Learning is Singular, and That's Good

arXiv · PDF · Open in the Atlas

Code that ran

13 of 17 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; 1 honoured the contract we drafted, 0 violated it and 12 have no contract verdict; 0 failures of our instrument; 4 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/2010.11560.json · badge: /api/ran/2010.11560.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
susanwe/RLCT canonical (classifier label)
classifier basis: census
13 ran · 4 no recorded run
FunctionStatusWhere it lives
conditioned_pyro_tanh Ran
ran on a synthesized input
susanwe/RLCT/models.py
code served (permissive licence) · get_code("d5e823409fd99f99")
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 Tensor, ndim 2, dtype family float, shape (2, 1) 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 pyro
import pyro.distributions as dist
import pyro.poutine as poutine
import numpy as np

torch.manual_seed(0)

def pyro_tanh(X, D_H, beta, Bmap=None, Amap=None):
    D_X = X.shape[1]
    if Amap is not None:
        A = pyro.sample('A', dist.Normal(Amap, torch.ones((D_X, D_H))))
    else:
        A = pyro.sample('A', dist.Normal(torch.zeros((D_X, D_H)), torch.ones((D_X, D_H))))
    z1 = torch.tanh(torch.matmul(X, A))
    if Bmap is not None:
        B = pyro.sample('B', dist.Normal(Bmap, torch.ones((D_H, 1))))
    else:
        B = pyro.sample('B', dist.Normal(torch.zeros((D_H, 1)), torch.ones((D_H, 1))))
    z2 = torch.matmul(z1, B)
    return pyro.sample('Y', dist.Normal(z2, 1 / np.sqrt(beta)))

def conditioned_pyro_tanh(pyro_tanh, X, Y, D_H, beta, Bmap, Amap):
    return poutine.condition(pyro_tanh, data={'Y': Y})(X, D_H, beta, Bmap, Amap)

X = torch.randn(2, 4)
Y = torch.randn(2, 1)
D_H = 3
beta = 1.0
Amap = torch.randn(4, 3)
Bmap = torch.randn(3, 1)

ARGS = {
    'pyro_tanh': pyro_tanh,
    'X': X,
    'Y': Y,
    'D_H': D_H,
    'beta': beta,
    'Bmap': Bmap,
    'Amap': Amap
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pyro, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 1) — shape (2, 1) 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
count_parameters Ran
ran, honoured the contract we drafted
susanwe/RLCT/RLCT_helper.py
code served (permissive licence) · get_code("f6b944f50d3f15ae")
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 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
import torch.nn as nn

torch.manual_seed(0)

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):
        x = self.linear1(x)
        x = self.linear2(x)
        return x

model = SimpleModel()

ARGS = {
    "model": model
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
expected_nll_posterior_relu Ran
ran on a synthesized input
susanwe/RLCT/mcmc_helper.py
code served (permissive licence) · get_code("dd81cf51ce199054")
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 torch
import pyro.distributions as dist

torch.manual_seed(0)

# Build a small args object with num_samples attribute
class Args:
    def __init__(self, num_samples):
        self.num_samples = num_samples

# Create small tensors for input and output
# X: batch_size=2, input_dim=4
X = torch.randn(2, 4)
# Y: batch_size=2, output_dim=1
Y = torch.randn(2, 1)

# Build samples dict with weight and bias tensors for a 5-layer network
# Layer shapes: 4 -> 3 -> 5 -> 6 -> 7 -> 1
num_samples = 3

samples = {
    'w1': torch.randn(num_samples, 4, 3),  # input_dim=4, hidden1=3
    'b1': torch.randn(num_samples, 3),
    'w2': torch.randn(num_samples, 3, 5),  # hidden1=3, hidden2=5
    'b2': torch.randn(num_samples, 5),
    'w3': torch.randn(num_samples, 5, 6),  # hidden2=5, hidden3=6
    'b3': torch.randn(num_samples, 6),
    'w4': torch.randn(num_samples, 6, 7),  # hidden3=6, hidden4=7
    'b4': torch.randn(num_samples, 7),
    'w5': torch.randn(num_samples, 7, 1),  # hidden4=7, output=1
    'b5': torch.randn(num_samples, 1),
}

args = Args(num_samples=num_samples)

ARGS = {
    'samples': samples,
    'args': 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
pyro, 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
expected_nll_posterior_tanh Ran
ran on a synthesized input
susanwe/RLCT/mcmc_helper.py
code served (permissive licence) · get_code("33119372a53640d5")
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 torch
import pyro.distributions as dist

torch.manual_seed(0)

class Args:
    def __init__(self, num_samples):
        self.num_samples = num_samples

samples = {
    'w': [torch.randn(3, 1) for _ in range(5)],
    'q': [torch.randn(1, 2) for _ in range(5)]
}

X = torch.randn(2, 3)
Y = torch.randn(2, 2)

args = Args(num_samples=5)

ARGS = {
    'samples': samples,
    'args': 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
pyro, 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
get_data Ran
ran on a synthesized input
susanwe/RLCT/langevin_monte_carlo.py
code served (permissive licence) · get_code("6ad62680134e71a5")
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 torch
import torch.nn as nn
from torch.distributions.multivariate_normal import MultivariateNormal
from torch.distributions.normal import Normal
from torch.distributions.uniform import Uniform

torch.manual_seed(0)

class Args:
    def __init__(self):
        self.dataset = 'lr'
        self.H = 4
        self.n = 2
        self.tanh = [-1.0, 1.0, 0.5]
        self.tanh_nontrivial = [4, 3, 5, 0.5, 0.1]
        self.rr = [4, 3, 5]
        self.w0 = None
        self.b0 = None
        self.y_std = None
        self.x_std = None
        self.H0 = None
        self.input_dim = None
        self.output_dim = None

ARGS = {
    'args': Args()
}
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 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
get_data_symmetric Ran
ran on a synthesized input
susanwe/RLCT/pyro_example.py
code served (permissive licence) · get_code("c669f2e8fa1858ea")
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 torch
import numpy as np

torch.manual_seed(0)

class Args:
    def __init__(self):
        self.num_data = 2
        self.symmetry_factor = 2
        self.num_hidden = 4

ARGS = {
    'args': Args()
}
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 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
lsfit_lambda Ran
ran on a synthesized input
susanwe/RLCT/RLCT_helper.py
code served (permissive licence) · get_code("5dc12c699e81fa9f")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
import torch
from unittest.mock import MagicMock

torch.manual_seed(0)

class ArgsConfig:
    def __init__(self):
        self.numbetas = 5
        self.betas = np.array([0.5, 1.0, 1.5, 2.0, 2.5])
        self.elasticnet_alpha = 0.01
        self.w_dim = 8
        self.trueRLCT = 2.5
        self.posterior_method = "VI"

args = ArgsConfig()

temperedNLL_perMC_perBeta = np.array([10.5, 9.8, 8.2, 7.1, 6.3])

saveimgname = None

ARGS = {
    "temperedNLL_perMC_perBeta": temperedNLL_perMC_perBeta,
    "args": args,
    "saveimgname": saveimgname
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, sklearn, statsmodels, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
nonlin Ran
ran on a synthesized input
susanwe/RLCT/pyro_example.py
code served (permissive licence) · get_code("fac793eb17aba0d5")
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 Tensor, ndim 2, dtype family float, shape (2, 4) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

# The function `nonlin` takes a single parameter `x` and applies torch.relu to it
# x should be a tensor
x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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 Tensor, ndim 2, dtype family float, shape (2, 4) — shape (2, 4) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
plot_energy Ran
ran on a synthesized input
susanwe/RLCT/langevin_monte_carlo.py
code served (permissive licence) · get_code("57da358ba6236f52")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np
from unittest.mock import MagicMock

# Create a mock args object with required attributes
args = MagicMock()
args.w_dim = 8
args.trueRLCT = 2.5
args.RLCT_ub = None
args.taskid = 1

# Create inverse_temp as a 1D numpy array
inverse_temp = np.array([0.5, 1.0, 1.5, 2.0, 2.5])

# Create avg_nlls_beta as a 1D numpy array matching the size of inverse_temp
avg_nlls_beta = np.array([10.5, 8.2, 6.8, 5.5, 4.3])

# Savefigname for the plot
savefigname = "energy_plot"

ARGS = {
    "args": args,
    "inverse_temp": inverse_temp,
    "avg_nlls_beta": avg_nlls_beta,
    "savefigname": savefigname,
    "savefig": False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
matplotlib, numpy, sklearn — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, ndim 1, dtype family float — a tuple output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
pyro_rr Ran
ran on a synthesized input
susanwe/RLCT/models.py
code served (permissive licence) · get_code("66467be486ae22c3")
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 Tensor, ndim 2, dtype family float, shape (2, 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 torch
import numpy as np

torch.manual_seed(0)

# Create small tensors with distinct dimensions
# X: (batch_size=2, D_X=4)
X = torch.randn(2, 4)

# Y: (batch_size=2, D_Y=3)
Y = torch.randn(2, 3)

# Hidden dimension
D_H = 5

# Beta parameter (regularization strength)
beta = 1.0

# Optional prior maps for A and B matrices
Amap = torch.randn(4, 5)  # (D_X, D_H)
Bmap = torch.randn(5, 3)  # (D_H, D_Y)

ARGS = {
    'X': X,
    'Y': Y,
    'D_H': D_H,
    'beta': beta,
    'Bmap': Bmap,
    'Amap': Amap,
    'relu': True,
    'lastlayeronly': False
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pyro, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 3) — shape (2, 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
pyro_tanh Ran
ran on a synthesized input
susanwe/RLCT/models.py
code served (permissive licence) · get_code("a6730396985a56d1")
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 Tensor, ndim 2, dtype family float, shape (2, 1) 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)

# Parameters for pyro_tanh
X = torch.randn(2, 4)  # batch size 2, D_X = 4
D_H = 3  # hidden dimension (distinct from D_X)
beta = 2.0  # inverse noise variance
Amap = torch.randn(4, 3)  # prior mean for A, shape (D_X, D_H)
Bmap = torch.randn(3, 1)  # prior mean for B, shape (D_H, 1)

ARGS = {
    'X': X,
    'D_H': D_H,
    'beta': beta,
    'Bmap': Bmap,
    'Amap': Amap,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pyro, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 2, dtype family float, shape (2, 1) — shape (2, 1) 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
run_inference Ran
ran on a synthesized input
susanwe/RLCT/pyro_example.py
code served (permissive licence) · get_code("ca8547ea3ea72bef")
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 MCMC. 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
import pyro
import pyro.distributions as dist
from pyro.infer import HMC, MCMC, NUTS

torch.manual_seed(0)

def nonlin(x):
    return torch.relu(x)

def model(X, Y, D_H, beta):
    D_X, D_Y = (X.shape[1], 1)
    w = pyro.sample('w', dist.Normal(torch.zeros((D_X, D_H)), torch.ones((D_X, D_H))))
    b = pyro.sample('b', dist.Normal(torch.zeros((1, D_H)), torch.ones((1, D_H))))
    z1 = nonlin(torch.matmul(X, w) + b)
    q = pyro.sample('q', dist.Normal(torch.zeros((D_H, 1)), torch.ones((D_H, 1))))
    c = pyro.sample('c', dist.Normal(torch.zeros((1, 1)), torch.ones((1, 1))))
    z2 = torch.matmul(z1, q) + c
    pyro.sample('Y', dist.Normal(z2, 1 / np.sqrt(beta)), obs=Y)

def run_inference(model, args, X, Y, beta):
    D_H = args.num_hidden
    start = time.time()
    kernel = NUTS(model, adapt_step_size=True)
    mcmc = MCMC(kernel, num_samples=args.num_samples, warmup_steps=args.num_warmup)
    mcmc.run(X, Y, D_H, beta)
    print('\nMCMC elapsed time:', time.time() - start)
    return mcmc

import time

class ArgsConfig:
    def __init__(self, num_hidden, num_samples, num_warmup):
        self.num_hidden = num_hidden
        self.num_samples = num_samples
        self.num_warmup = num_warmup

X = torch.randn(2, 4)
Y = torch.randn(2, 1)
beta = 1.0

args = ArgsConfig(num_hidden=3, num_samples=2, num_warmup=1)

ARGS = {
    'model': model,
    'args': args,
    'X': X,
    'Y': Y,
    'beta': beta
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pyro, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type MCMC — a MCMC output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a MCMC 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
theoretical_RLCT Ran
ran on a synthesized input
susanwe/RLCT/RLCT_helper.py
code served (permissive licence) · get_code("2433cc567caefb30")
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 torch.nn as nn

torch.manual_seed(0)

# The function only uses the 'model' parameter as a string comparison ('rr')
# and unpacks params into 4 integer values: input_dim, output_dim, H0, H
# No actual model object or tensor operations are performed in the function body

# Create valid parameters that satisfy the conditions for a meaningful RLCT calculation
input_dim = 5
output_dim = 4
H0 = 2
H = 3

# Verify the conditions are met:
# cond1: output_dim + H0 <= input_dim + H => 4 + 2 <= 5 + 3 => 6 <= 8 ✓
# cond2: input_dim + H0 <= output_dim + H => 5 + 2 <= 4 + 3 => 7 <= 7 ✓
# cond3: H + H0 <= input_dim + output_dim => 3 + 2 <= 5 + 4 => 5 <= 9 ✓

ARGS = {
    'model': 'rr',
    'params': [input_dim, output_dim, H0, H]
}
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
approxinf_nll_implicit Not yet run
unverified (harvested, no recorded run)
susanwe/RLCT/implicit_vi.py
code served (permissive licence) · get_code("057d9f485a882881")
How it ran

no invocation recorded.

get_dataset_by_id Not yet run
unverified (harvested, no recorded run)
susanwe/RLCT/dataset_factory.py
code served (permissive licence) · get_code("b698e60b64037283")
How it ran

no invocation recorded.

sample_EVI Not yet run
unverified (harvested, no recorded run)
susanwe/RLCT/explicit_vi.py
code served (permissive licence) · get_code("18d1e4cee617e2f6")
How it ran

no invocation recorded.

sample_IVI Not yet run
unverified (harvested, no recorded run)
susanwe/RLCT/implicit_vi.py
code served (permissive licence) · get_code("d386ff8d9a8246b5")
How it ran

no invocation recorded.

Also present in other papers' code

Of 17 samples lifted from this paper's linked repositories, 17 carry a function identity and 0 do not. Of the 17 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 1 the same month, 1 later) and 16 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
count_parameters 180 other papers
26 dated earlier, 2 the same month, 152 later · 8 only as an assembly brick
1705.08415 (2017-05)
Supervised Community Detection with Line Graph Neural Networks
1707.01926 (2017-07)
Diffusion Convolutional Recurrent Neural Network: Data-Driven Traffic Forecasting
1808.09367 (2018-08)
Deriving Machine Attention from Human Rationales
and 177 more not named here

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

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

Abstract

In singular models, the optimal set of parameters forms an analytic set with singularities and classical statistical inference cannot be applied to such models. This is significant for deep learning as neural networks are singular and thus "dividing" by the determinant of the Hessian or employing the Laplace approximation are not appropriate. Despite its potential for addressing fundamental issues in deep learning, singular learning theory appears to have made little inroads into the developing canon of deep learning theory. Via a mix of theory and experiment, we present an invitation to singular learning theory as a vehicle for understanding deep learning and suggest important future work to make singular learning theory directly applicable to how deep learning is performed in practice.

For agents

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

get_harvested_code_for_paper("2010.11560")
get_code_for_paper("2010.11560")
have("2010.11560")

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

curl https://syntology.ai/api/ran/2010.11560.json

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

[![Syntology run record](https://syntology.ai/api/ran/2010.11560.svg)](https://syntology.ai/paper/2010.11560)

Connect an agent — have() is free.