SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2111.03383 · 2021

A Bayesian generative neural network framework for epidemic inference problems

arXiv · PDF · Open in the Atlas

Code that ran

19 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 19 have no contract verdict; 0 failures of our instrument; 5 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/2111.03383.json · badge: /api/ran/2111.03383.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
ocadni/annfore canonical (classifier label)
classifier basis not recorded on the link
19 ran · 5 no recorded run
FunctionStatusWhere it lives
batch_mm Ran
ran on a synthesized input
ocadni/annfore/annfore/models/common.py
code served (permissive licence) · get_code("88a25afdb229a4de")
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 3, dtype family float, shape (2, 5, 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

torch.manual_seed(0)

# Construct arguments for batch_mm
# matrix: Sparse or dense matrix, size (m, n)
# matrix_batch: Batched dense matrices, size (b, n, k)

# Let's use:
# m = 5 (output rows)
# n = 4 (shared dimension)
# k = 3 (output columns)
# b = 2 (batch size)

m, n, k, b = 5, 4, 3, 2

# Create a dense matrix of size (m, n) = (5, 4)
matrix = torch.randn(m, n)

# Create batched dense matrices of size (b, n, k) = (2, 4, 3)
matrix_batch = torch.randn(b, n, k)

ARGS = {
    "matrix": matrix,
    "matrix_batch": matrix_batch,
}
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 3, dtype family float, shape (2, 5, 3) — shape (2, 5, 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
calc_avg_states Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/l_utils.py
code served (permissive licence) · get_code("e46492b8d6cb0304")
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 torch

torch.manual_seed(0)

# The function expects:
# - samples: a tensor with shape (batch, time, spatial, features)
#   where it indexes samples[:, T, :, 0], samples[:, T, :, 1], samples[:, T, :, 2]
#   So we need at least 3 feature channels (S, I, R)
# - T: an integer time index (default 0)

# Create a small samples tensor
# Shape: (batch_size=2, time_steps=5, spatial_dim=4, features=3)
samples = torch.randn(2, 5, 4, 3)

# T is already defaulted to 0 in the function, but we can specify it
T = 0

ARGS = {
    "samples": samples,
    "T": T,
}
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, 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
calc_feat_power Ran
ran on a synthesized input
ocadni/annfore/annfore/net/deep_linear.py
code served (permissive licence) · get_code("f32d4ccb471ff352")
How it ran

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

import numpy as np

ARGS = {
    "dim_input": 100,
    "out_count": 10,
    "nlay": 5,
    "power": 2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family int — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
format_src_print Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/l_utils.py
code served (permissive licence) · get_code("8e6aaa40cc72bcbb")
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 format_src_print accepts a single parameter 'sourc'
# It can handle: torch.Tensor, np.ndarray, or list-like objects
# The function indexes x[i] and accesses x[i][0] and x[i][1], suggesting 2D array
# x.shape[0] is used, so we need a 2D structure with at least 1 row

# Create a small 2D numpy array (2 rows, 2 columns for the unpacking)
sourc_numpy = np.array([[5, 0.123], [3, 0.456]])

ARGS = {
    "sourc": sourc_numpy
}
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
logP Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/train_params.py
code served (permissive licence) · get_code("db96eb179ca980fe")
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 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)

class EnergyModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def energy(self, x):
        return self.linear(x).squeeze(-1)

model = EnergyModel()
x = torch.randn(2, 4)

ARGS = {
    "model": model,
    "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 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
logP_logQ Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/train_params.py
code served (permissive licence) · get_code("33352cf70ab668d7")
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 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)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
        self.extra_params = {'max_Z': -200.0}
    
    def energy(self, x):
        return self.linear(x).squeeze(-1)

model = SimpleModel()
x = torch.randn(2, 4)
log_q_x = torch.randn(2)
baseline_norm = -200

ARGS = {
    'model': model,
    'x': x,
    'log_q_x': log_q_x,
    'baseline_norm': baseline_norm
}
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 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
loss_fn Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/losses.py
code served (permissive licence) · get_code("8260a897032b015e")
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 function. 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 time

torch.manual_seed(0)

# Capability enum
class Capabil:
    ENERGY_SEP = "energy_separated"

# Minimal net model that satisfies the function's usage
class MinimalNet(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def sample(self, num_samples):
        # Returns (samples, probs)
        samples = torch.randn(num_samples, 4)
        probs = torch.ones(num_samples, 4) * 0.25
        return samples, probs
    
    def __call__(self, samples):
        # Returns log probabilities
        return torch.ones(samples.shape[0], samples.shape[1]) * -1.386
    
    def _log_prob(self, samples, probs):
        # Returns log probability per sample
        return torch.sum(torch.log(probs + 1e-8), dim=1)
    
    def transform_samples(self, samples):
        # Identity transform
        return samples

# Minimal model that provides energy computation
class MinimalEnergyModel(nn.Module):
    def __init__(self, separate_energies=True):
        super().__init__()
        self.linear = nn.Linear(4, 1)
        self.separate_energies = separate_energies
        self._capabilities = set()
        if separate_energies:
            self._capabilities.add(Capabil.ENERGY_SEP)
    
    def get_capabilities(self):
        return self._capabilities
    
    def energy_separated(self, samples):
        # Returns 4 energy tensors (one per ENERGIES_NAMES entry)
        batch_size = samples.shape[0]
        e1 = torch.randn(batch_size)
        e2 = torch.randn(batch_size)
        e3 = torch.randn(batch_size)
        e4 = torch.randn(batch_size)
        return [e1, e2, e3, e4]
    
    def energy_(self, samples):
        # Returns single energy tensor
        batch_size = samples.shape[0]
        return torch.randn(batch_size)

# Build minimal instances
net = MinimalNet()
model = MinimalEnergyModel(separate_energies=True)
beta = 1.0

ARGS = {
    'net': net,
    'model': model,
    'beta': beta,
    'rerun_probs': 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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
loss_fn_coeff Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/losses.py
code served (permissive licence) · get_code("a7f1d520148c4025")
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 function. 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 time

torch.manual_seed(0)

# Define Capabil enum or class for capabilities
class Capabil:
    ENERGY_SEP = "energy_separated"

# Define a minimal net model that implements required methods
class MinimalNet(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def sample(self, num_samples):
        # Returns (samples, probs)
        samples = torch.randn(num_samples, 4)
        probs = torch.softmax(self.linear(samples), dim=-1)
        return samples, probs
    
    def _log_prob(self, samples, probs):
        # Return log probabilities for samples
        return torch.sum(torch.log(probs + 1e-8), dim=-1)
    
    def transform_samples(self, samples):
        # Transform samples (identity in minimal case)
        return samples

# Define a minimal model that implements energy methods
class MinimalEnergyModel(nn.Module):
    def __init__(self, has_separated=True):
        super().__init__()
        self.linear = nn.Linear(4, 1)
        self.has_separated = has_separated
        self._capabilities = [Capabil.ENERGY_SEP] if has_separated else []
    
    def get_capabilities(self):
        return self._capabilities
    
    def energy_separated(self, samples):
        # Return 4 separate energy tensors for ENERGIES_NAMES
        energy = self.linear(samples).squeeze(-1)
        return [
            energy.abs(),
            energy.abs() * 0.5,
            energy.abs() * 0.3,
            energy.abs() * 0.2
        ]
    
    def energy_(self, samples):
        # Return combined energy
        energy = self.linear(samples).squeeze(-1)
        return energy.abs()

# Create instances
net = MinimalNet()
model = MinimalEnergyModel(has_separated=True)
beta = 0.5

ARGS = {
    'net': net,
    'model': model,
    'beta': beta
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
loss_fn_coeff_p_sus Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/losses.py
code served (permissive licence) · get_code("22fae26097751c91")
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 function. 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 time

torch.manual_seed(0)

# Create a mock network that implements the required methods
class MockNet(nn.Module):
    def __init__(self, sample_dim=4, latent_dim=3):
        super().__init__()
        self.sample_dim = sample_dim
        self.latent_dim = latent_dim
        self.linear = nn.Linear(latent_dim, sample_dim)
    
    def sample(self, num_samples):
        # Returns (samples, probs)
        samples = torch.randn(num_samples, self.sample_dim)
        probs = torch.rand(num_samples, self.sample_dim)
        return samples, probs
    
    def _log_prob(self, samples, probs):
        # Returns log probability for each sample
        return torch.randn(samples.shape[0])
    
    def transform_samples(self, samples):
        # Identity transform for simplicity
        return samples

# Create a mock model that implements energy_separated
class MockModel(nn.Module):
    def __init__(self, sample_dim=4):
        super().__init__()
        self.sample_dim = sample_dim
    
    def energy_separated(self, samples):
        # Returns tuple of 4 energy tensors (ener_mc, ener_obs, ener_sources, ener_p_sus)
        batch_size = samples.shape[0]
        ener_mc = torch.randn(batch_size)
        ener_obs = torch.randn(batch_size)
        ener_sources = torch.randn(batch_size)
        ener_p_sus = torch.randn(batch_size)
        return (ener_mc, ener_obs, ener_sources, ener_p_sus)

# Create instances
net = MockNet(sample_dim=4, latent_dim=3)
model = MockModel(sample_dim=4)

# Define beta parameter
beta = 0.5

# Optional fun_beta_inv parameter
def fun_beta_inv(b):
    return 1.0 - b

ARGS = {
    'net': net,
    'model': model,
    'beta': beta,
    'fun_beta_inv': fun_beta_inv
}
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 function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
make_lin_layers Ran
ran on a synthesized input
ocadni/annfore/annfore/net/deep_linear.py
code served (permissive licence) · get_code("70402e4f401c4103")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family 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 torch
import torch.nn as nn

torch.manual_seed(0)

# Create simple activation functions to use as in_func and last_func
in_func = nn.ReLU()
last_func = nn.Sigmoid()

# Define the features list - this determines layer dimensions
# The function iterates through features[:-1] and connects each to the next
# So features = [3, 5] means: input layer with 3 features, output layer with 5 features
features = [3, 5]

# bias parameter
bias = True

ARGS = {
    'features': features,
    'in_func': in_func,
    'last_func': last_func,
    'bias': bias
}
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 list, ndim 1, 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
make_masks Ran
ran on a synthesized input
ocadni/annfore/annfore/net/nn_sir_path_obs.py
code served (permissive licence) · get_code("88747e21714a2d36")
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 3, dtype family float, shape (5, 2, 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

# Construct observations list
# Each observation is [node_index, value, time]
# value: 0 (susceptible), 1 (infected), 2 (recovered)
observ = [
    [0, 0, 1],      # node 0, susceptible at time 1
    [1, 1, 2],      # node 1, infected at time 2
    [2, 2, 3],      # node 2, recovered at time 3
]

# Parameters
n = 5               # number of nodes
T = 7               # number of time steps (times range from 0 to T inclusive)

ARGS = {
    "observ": observ,
    "n": n,
    "T": T,
}
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 3, dtype family float, shape (5, 2, 8) — shape (5, 2, 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
make_opt Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/opt.py
code served (permissive licence) · get_code("f4f47a0dd89c33fc")
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 Adam. 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)

# Create a simple neural network with parameters
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

model = SimpleModel()
params = model.parameters()

ARGS = {
    'params': params,
    'lr': 0.001,
    'opt': 'adam',
    'alpha': 0.99,
    'momentum': 0.9,
    'betas': (0.9, 0.999),
    'weight_decay': 0
}
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 — a Adam output has no array shape of its own; none 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
my_linear Ran
ran on a synthesized input
ocadni/annfore/annfore/net/deep_linear.py
code served (permissive licence) · get_code("30a691fc43d7825e")
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 Linear. 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.nn.parameter import Parameter
from torch.nn import init

torch.manual_seed(0)

class ZeroLinear(nn.Module):

    def __init__(self, out_features, bias):
        super(ZeroLinear, self).__init__()
        self.out_features = out_features
        self.in_features = 0
        if bias:
            self.bias = Parameter(torch.Tensor(out_features))
            self.has_bias = True
        else:
            self.bias = torch.zeros(out_features)
            self.has_bias = False
        self.reset_parameters()

    def reset_parameters(self, rang=0.1):
        if self.has_bias:
            init.uniform_(self.bias, -rang, rang)

    def forward(self, x):
        return self.bias.repeat(x.shape[0], 1)

    def extra_repr(self):
        return 'in_features=0, out_features={}, bias={}'.format(self.out_features, self.bias is not None)

def my_linear(in_feat, out_feat, bias):
    if in_feat > 0:
        return nn.Linear(in_feat, out_feat, bias)
    else:
        return ZeroLinear(out_feat, bias)

ARGS = {
    "in_feat": 5,
    "out_feat": 3,
    "bias": 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 Linear — a Linear output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Linear 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
normalize Ran
ran on a synthesized input
ocadni/annfore/annfore/models/sir_model_sep.py
code served (permissive licence) · get_code("90ce614d4b6b5793")
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 normalize takes a tensor and an optional dim parameter
# It calls .float() and .sum() on the tensor
# So we need a real torch tensor

tensor = torch.tensor([[1, 2, 3, 4],
                       [5, 6, 7, 8]], dtype=torch.int32)

ARGS = {
    "tensor": tensor,
    "dim": None
}
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
one_hot_conf_from_times Ran
ran on a synthesized input
ocadni/annfore/annfore/utils/conf_utils.py
code served (permissive licence) · get_code("23f65953c94eee38")
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 4, dtype family float, shape (2, 4, 5, 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

torch.manual_seed(0)

# x is a 3D tensor with shape (batch, nodes, 2)
# where the last dimension contains [infection_time, recovery_time]
# T is the number of time steps
batch_size = 2
num_nodes = 4
num_time_steps = 5

x = torch.tensor([
    [[1.0, 3.0], [0.0, 2.0], [2.0, 4.0], [1.0, 1.5]],
    [[0.0, 2.0], [1.0, 3.0], [0.5, 2.5], [2.0, 3.0]]
], dtype=torch.float32)

T = num_time_steps

ARGS = {
    "x": x,
    "T": T
}
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 4, dtype family float, shape (2, 4, 5, 3) — shape (2, 4, 5, 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
one_hot_samples Ran
ran on a synthesized input
ocadni/annfore/annfore/utils/conf_utils.py
code served (permissive licence) · get_code("3eb2444e97ae7693")
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 4, dtype family float, shape (2, 4, 5, 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

torch.manual_seed(0)

# Construct minimal valid arguments for one_hot_samples
# The function expects:
# - x: tensor of shape (batch_size, num_features, 2) where x[:, :, 0] is t_i and x[:, :, 1] is t_r
# - T: integer representing number of time steps
# - q: integer representing number of one-hot classes (default 3)

batch_size = 2
num_features = 4
x = torch.randn(batch_size, num_features, 2)

T = 5
q = 3

ARGS = {
    "x": x,
    "T": T,
    "q": q,
}
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 4, dtype family float, shape (2, 4, 5, 3) — shape (2, 4, 5, 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
one_hot_samples_relative_r Ran
ran on a synthesized input
ocadni/annfore/annfore/utils/conf_utils.py
code served (permissive licence) · get_code("b326d2cdb16132fc")
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 4, dtype family float, shape (2, 4, 5, 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

torch.manual_seed(0)

# Analyze the function:
# - x: tensor with shape (batch, dim, 2) where x[:,:,0] is t_i and x[:,:,1] is t_r
# - T: integer for time dimension
# - q: integer for number of one-hot categories (default 3)
# - Returns tensor of shape (batch, dim, T, q)

batch_size = 2
dim = 4
T = 5
q = 3

# Create x tensor with shape (batch_size, dim, 2)
# x[:, :, 0] contains t_i (initial times)
# x[:, :, 1] contains t_r (relative times/durations)
x = torch.randint(0, T, (batch_size, dim, 2), dtype=torch.float32)

ARGS = {
    'x': x,
    'T': T,
    'q': q,
}
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 4, dtype family float, shape (2, 4, 5, 3) — shape (2, 4, 5, 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
sample_avg_states Ran
ran on a synthesized input
ocadni/annfore/annfore/learn/l_utils.py
code served (permissive licence) · get_code("8de71585e15ab1b8")
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 torch
import torch.nn as nn

torch.manual_seed(0)

class MockNet(nn.Module):
    def __init__(self):
        super().__init__()
    
    def sample(self, sample_size):
        # Returns (sample_raw, x_hat)
        # sample_raw needs to be transformable by transform_samples
        # Let's create a batch of samples
        sample_raw = torch.randn(sample_size, 10, 3)  # sample_size x time_steps x features
        x_hat = torch.randn(sample_size, 3)
        return sample_raw, x_hat
    
    def transform_samples(self, sample_raw):
        # Takes sample_raw and returns samples with shape (batch, time, spatial, state_dim)
        # calc_avg_states expects samples[:, T, :, 0], samples[:, T, :, 1], samples[:, T, :, 2]
        # So we need shape: (batch_size, time_steps, spatial_dims, 3)
        batch_size = sample_raw.shape[0]
        time_steps = sample_raw.shape[1]
        # Create a 4D tensor: (batch, time, spatial, state_vars)
        samples = torch.randn(batch_size, time_steps, 4, 3)
        return samples

net = MockNet()
sample_size = 2
T = 0

ARGS = {
    "sample_size": sample_size,
    "net": net,
    "T": T
}
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, 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
transform_to_spins Ran
ran on a synthesized input
ocadni/annfore/annfore/models/sir_model_sep.py
code served (permissive licence) · get_code("9fb5bf91b56bc932")
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)

# Create a small binary tensor (batch_size=2, dim=4)
bin_tensor = torch.tensor([
    [0, 1, 0, 1],
    [1, 0, 1, 0]
], dtype=torch.float32)

ARGS = {
    'bin_tensor': bin_tensor
}
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
calc_p_no_c_np Not yet run
unverified (harvested, no recorded run)
ocadni/annfore/annfore/models/common.py
code served (permissive licence) · get_code("d30029f478a818ae")
How it ran

no invocation recorded.

check_convert_tensor Not yet run
unverified (harvested, no recorded run)
ocadni/annfore/annfore/models/common.py
code served (permissive licence) · get_code("169404d17a6af306")
How it ran

no invocation recorded.

check_convert_tensor Not yet run
unverified (harvested, no recorded run)
ocadni/annfore/annfore/models/sir_model_sep.py
code served (permissive licence) · get_code("7010edca19a563be")
How it ran

no invocation recorded.

get_linear_layers_path Not yet run
unverified (harvested, no recorded run)
ocadni/annfore/annfore/net/base.py
code served (permissive licence) · get_code("5e93c65f419d5914")
How it ran

no invocation recorded.

opt_param_init Not yet run
unverified (harvested, no recorded run)
ocadni/annfore/annfore/learn/train_params.py
code served (permissive licence) · get_code("25fc7691e23186cf")
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 24 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 24 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 reconstruction of missing information in epidemic spreading on contact networks can be essential in the prevention and containment strategies. The identification and warning of infectious but asymptomatic individuals (i.e., contact tracing), the well-known patient-zero problem, or the inference of the infectivity values in structured populations are examples of significant epidemic inference problems. As the number of possible epidemic cascades grows exponentially with the number of individuals involved and only an almost negligible subset of them is compatible with the observations (e.g., medical tests), epidemic inference in contact networks poses incredible computational challenges. We present a new generative neural networks framework that learns to generate the most probable infection cascades compatible with observations. The proposed method achieves better (in some cases, significantly better) or comparable results with existing methods in all problems considered both in synthetic and real contact networks. Given its generality, clear Bayesian and variational nature, the presented framework paves the way to solve fundamental inference epidemic problems with high precision in small and medium-sized real case scenarios such as the spread of infections in workplaces and hospitals.

For agents

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

get_harvested_code_for_paper("2111.03383")
get_code_for_paper("2111.03383")
have("2111.03383")

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

curl https://syntology.ai/api/ran/2111.03383.json

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

[![Syntology run record](https://syntology.ai/api/ran/2111.03383.svg)](https://syntology.ai/paper/2111.03383)

Connect an agent — have() is free.