SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2605.29933 · 2026

CLUBench: A Clustering Benchmark

Jicong Fan, Chris Ding, Dazhi Fu, Feng Xiao

arXiv · PDF · Open in the Atlas

Code that ran

13 of 15 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; 3 honoured the contract we drafted, 0 violated it and 3 have no contract verdict; 7 failures of our instrument; 2 no recorded run), measured 2026-11-22; 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/2605.29933.json · badge: /api/ran/2605.29933.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
xiaofeng-github/CLUBench canonical (classifier label)
classifier basis: census
13 ran · 2 no recorded run
FunctionStatusWhere it lives
SP_for_affine_matrix Ran
ran; our contract draft was wrong, not the code
xiaofeng-github/CLUBench/CLUBench/algorithms/DSCN.py
code served (permissive licence) · get_code("82ab575081aaa4a6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family int, shape (5,) 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
from sklearn.cluster import KMeans

torch.manual_seed(0)

# Create a small affinity/adjacency matrix A
# The function expects A to be a square tensor that represents a graph affinity matrix
# It should be symmetric and positive semi-definite for meaningful spectral clustering
n_nodes = 5
A = torch.tensor([
    [2.0, 0.5, 0.3, 0.1, 0.0],
    [0.5, 2.0, 0.4, 0.2, 0.1],
    [0.3, 0.4, 2.0, 0.5, 0.2],
    [0.1, 0.2, 0.5, 2.0, 0.6],
    [0.0, 0.1, 0.2, 0.6, 2.0]
], dtype=torch.float32)

# Ensure A is symmetric
A = (A + A.T) / 2

# Make it positive semi-definite by adding to diagonal
A = A + torch.eye(n_nodes) * 0.5

n_clusters = 2

ARGS = {
    "n_clusters": n_clusters,
    "A": A
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family int, shape (5,) — shape (5,) 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
cluster_accuracy Ran
ran; our contract draft was wrong, not the code
xiaofeng-github/CLUBench/CLUBench/algorithms/DEC.py
code served (permissive licence) · get_code("139f2b700934ddfc")
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

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

# Create small sample data for clustering
# y_true: ground truth cluster assignments
y_true = np.array([0, 0, 1, 1, 2, 2], dtype=np.int64)

# y_predicted: predicted cluster assignments (may be in different order)
y_predicted = np.array([1, 1, 0, 0, 2, 2], dtype=np.int64)

# cluster_number: number of clusters (3 in this case)
cluster_number = 3

ARGS = {
    'y_true': y_true,
    'y_predicted': y_predicted,
    'cluster_number': cluster_number
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple, 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
compute_joint Ran
ran; our contract draft was wrong, not the code
xiaofeng-github/CLUBench/CLUBench/algorithms/DMICC.py
code served (permissive licence) · get_code("a1a92236d08b7517")
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 (4, 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)

# view1 and view2 are 2D tensors of shape (batch_size, k)
# The function performs operations:
# - unsqueeze to get (bn, k, 1) and (bn, 1, k)
# - element-wise multiply
# - sum over batch dimension dim=0
# - transpose, add, divide
# - normalize by sum

bn = 2  # batch size
k = 4   # feature dimension

view1 = torch.randn(bn, k)
view2 = torch.randn(bn, k)

ARGS = {
    "view1": view1,
    "view2": view2,
}
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 (4, 4) — shape (4, 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
crossview_contrastive_Loss Ran
ran; our contract draft was wrong, not the code
xiaofeng-github/CLUBench/CLUBench/algorithms/DMICC.py
code served (permissive licence) · get_code("4dc209621c3bab5c")
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 sys

torch.manual_seed(0)

# Create small probability-like tensors
# view1 and view2 should have shape (batch_size, k)
# where batch_size is the number of samples and k is the embedding dimension
batch_size = 2
k = 4

# Create random tensors and normalize them to be probability-like (non-negative, sum to 1 per row)
view1_raw = torch.rand(batch_size, k)
view1 = view1_raw / view1_raw.sum(dim=1, keepdim=True)

view2_raw = torch.rand(batch_size, k)
view2 = view2_raw / view2_raw.sum(dim=1, keepdim=True)

ARGS = {
    "view1": view1,
    "view2": view2,
    "EPS": sys.float_info.epsilon
}
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
spectral_clustering Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
xiaofeng-github/CLUBench/CLUBench/algorithms/DSCN.py
code served (permissive licence) · get_code("58e3b24d4d477a66")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family int, shape (5,) 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
from sklearn.cluster import KMeans

torch.manual_seed(0)
np.random.seed(0)

# Create a small affinity matrix C (5x5)
# This should be symmetric or close to symmetric for the algorithm
n_samples = 5
C = np.random.rand(n_samples, n_samples)
C = 0.5 * (C + C.T)  # Make symmetric
C = np.abs(C)  # Ensure non-negative

# Parameters
K = 2  # number of clusters
d = 1  # dimension parameter for svds
alpha = 0.5  # threshold parameter (between 0 and 1)
ro = 2  # power parameter

ARGS = {
    "C": C,
    "K": K,
    "d": d,
    "alpha": alpha,
    "ro": ro
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, scipy, sklearn, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 1, dtype family int, shape (5,) — shape (5,) 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
target_distribution Ran
ran; our contract draft was wrong, not the code
xiaofeng-github/CLUBench/CLUBench/algorithms/DEC.py
code served (permissive licence) · get_code("cc8c969ec45f0b92")
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 batch tensor with shape [batch_size=2, num_clusters=4]
# The function performs operations: squaring, sum over dimension 0, transpose, and sum over dimension 1
# So we need a 2D tensor of float type
batch = torch.randn(2, 4, dtype=torch.float32)

# Ensure the batch has positive values since we're squaring and dividing
# (to avoid numerical issues with the target distribution)
batch = torch.abs(batch) + 0.1

ARGS = {
    "batch": 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 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
test Ran
ran, honoured the contract we drafted
xiaofeng-github/CLUBench/CLUBench/algorithms/CC.py
code served (permissive licence) · get_code("fb2661c1f273dcdf")
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 torch.nn as nn
from torch.utils.data import TensorDataset

torch.manual_seed(0)

class SimpleClusterModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
        self.device = 'cpu'
    
    def forward_cluster(self, x):
        return self.linear(x)

model = SimpleClusterModel()
model.eval()

x_data = torch.randn(2, 4)
y_data = torch.randint(0, 2, (2,))

dataset = TensorDataset(x_data, y_data)

batch_size = 2

ARGS = {
    'model': model,
    'dataset': dataset,
    'batch_size': batch_size
}
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, 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
test Ran
ran on a synthesized input
xiaofeng-github/CLUBench/CLUBench/algorithms/DIVC.py
code served (permissive licence) · get_code("cf35aad7d0c5fa37")
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 1, dtype family int, shape (6,) 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
from torch.utils.data import Dataset

torch.manual_seed(0)

# Define a minimal model with required attributes/methods
class MinimalModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
        self.device = torch.device('cpu')
    
    def forward(self, x):
        return self.linear(x)
    
    def predcit(self, x):
        # Note: function has typo "predcit" instead of "predict"
        return self.forward(x)
    
    def eval(self):
        super().eval()
        return self

# Define a minimal dataset
class MinimalDataset(Dataset):
    def __init__(self, size=6):
        self.size = size
    
    def __len__(self):
        return self.size
    
    def __getitem__(self, idx):
        x = torch.randn(4)
        y = torch.randint(0, 3, (1,)).item()
        return x, y

# Create instances
model = MinimalModel()
dataset = MinimalDataset(size=6)
batch_size = 2

ARGS = {
    'model': model,
    'dataset': dataset,
    'batch_size': batch_size
}
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 1, dtype family int, shape (6,) — shape (6,) 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
test Ran
ran on a synthesized input
xiaofeng-github/CLUBench/CLUBench/algorithms/PICA.py
code served (permissive licence) · get_code("5bbb107a05f6eb61")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 1, dtype family int, shape (4,) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from torch.utils.data import Dataset

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
        self.device = 'cpu'
    
    def forward(self, x):
        return self.linear(x)
    
    def predcit(self, x):
        return self.forward(x)

class SimpleDataset(Dataset):
    def __init__(self, num_samples=4):
        self.num_samples = num_samples
    
    def __len__(self):
        return self.num_samples
    
    def __getitem__(self, idx):
        x = torch.randn(4)
        y = torch.randint(0, 3, (1,))
        return x, y

model = SimpleModel()
dataset = SimpleDataset(num_samples=4)
batch_size = 2

ARGS = {
    'model': model,
    'dataset': dataset,
    'batch_size': batch_size
}
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 ndarray, ndim 1, dtype family int, shape (4,) — shape (4,) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
thrC Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
xiaofeng-github/CLUBench/CLUBench/algorithms/DSCN.py
code served (permissive licence) · get_code("0ee97af68ffc2ea5")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned ndarray, ndim 2, dtype family float, shape (5, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# Create a small correlation matrix C of shape (N, N)
# The function uses C.shape[1] to get N, and operates on columns
N = 5
C = np.array([
    [1.0, 0.8, -0.3, 0.5, 0.2],
    [0.8, 1.0, 0.4, -0.2, 0.6],
    [-0.3, 0.4, 1.0, 0.7, -0.1],
    [0.5, -0.2, 0.7, 1.0, 0.3],
    [0.2, 0.6, -0.1, 0.3, 1.0]
], dtype=np.float64)

# alpha controls the threshold; test with a value < 1
alpha = 0.5

ARGS = {
    "C": C,
    "alpha": alpha
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 2, dtype family float, shape (5, 5) — shape (5, 5) 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
train Ran
ran on a synthesized input
xiaofeng-github/CLUBench/CLUBench/algorithms/CC.py
code served (permissive licence) · get_code("d0f49064fbed326b")
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 3, dtype family float; no output shape is stated. 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 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 SimpleModel(nn.Module):
    def __init__(self, input_dim, output_dim, cluster_num):
        super(SimpleModel, self).__init__()
        self.device = 'cpu'
        self.cluster_num = cluster_num
        self.fc1 = nn.Linear(input_dim, 16)
        self.fc_z = nn.Linear(16, output_dim)
        self.fc_c = nn.Linear(16, cluster_num)
        
    def forward(self, x_i, x_j):
        h_i = torch.relu(self.fc1(x_i))
        h_j = torch.relu(self.fc1(x_j))
        z_i = self.fc_z(h_i)
        z_j = self.fc_z(h_j)
        c_i = torch.softmax(self.fc_c(h_i), dim=1)
        c_j = torch.softmax(self.fc_c(h_j), dim=1)
        return z_i, z_j, c_i, c_j
    
    def forward_cluster(self, x):
        h = torch.relu(self.fc1(x))
        c = torch.softmax(self.fc_c(h), dim=1)
        return c

class SimpleDataset:
    def __init__(self, num_samples, input_dim):
        self.num_samples = num_samples
        self.input_dim = input_dim
    
    def __len__(self):
        return self.num_samples
    
    def __getitem__(self, idx):
        x = torch.randn(self.input_dim)
        y = torch.tensor(0)
        return x, y

model = SimpleModel(input_dim=4, output_dim=8, cluster_num=3)
dataset = SimpleDataset(num_samples=10, input_dim=4)

ARGS = {
    'model': model,
    'dataset': dataset,
    'epochs': 2,
    'stop_epochs': 1,
    'noise_std': 0.1,
    'lr': 0.001,
    'batch_size': 2,
    'instance_temp': 0.5,
    'cluster_temp': 1.0,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 3, 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
train Ran
ran, honoured the contract we drafted
xiaofeng-github/CLUBench/CLUBench/algorithms/DIVC.py
code served (permissive licence) · get_code("47d9f496a3f5f2d3")
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 2, dtype family int; no output shape is stated. 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 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
from torch.utils.data import TensorDataset

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self, device='cpu'):
        super().__init__()
        self.device = device
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        # Returns a loss value (scalar tensor)
        output = self.linear(x)
        loss = output.mean()
        return loss
    
    def predcit(self, x):
        # Note: function has typo "predcit" instead of "predict"
        output = self.linear(x)
        return output

# Create a small dataset with 10 samples
# Each sample has 4 features and 1 label
X = torch.randn(10, 4)
y = torch.randint(0, 3, (10,))
dataset = TensorDataset(X, y)

# Define arguments for train function
ARGS = {
    'model': SimpleModel(device='cpu'),
    'dataset': dataset,
    'epochs': 2,
    'stop_epochs': 1,
    'lr': 0.001,
    'batch_size': 2,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family 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
train Ran
ran, honoured the contract we drafted
xiaofeng-github/CLUBench/CLUBench/algorithms/PICA.py
code served (permissive licence) · get_code("83c6fcc5dbf3b90f")
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 2, dtype family int; no output shape is stated. 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 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
from torch.utils.data import TensorDataset

torch.manual_seed(0)

class DummyModel(nn.Module):
    def __init__(self, device='cpu'):
        super().__init__()
        self.device = device
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        # Returns loss for training
        logits = self.linear(x)
        # Simple dummy loss
        loss = logits.mean()
        return loss
    
    def predcit(self, x):
        # Note: function has typo "predcit" instead of "predict"
        logits = self.linear(x)
        return logits

# Create a small dataset: 8 samples, 4 features, 3 classes
X = torch.randn(8, 4)
y = torch.randint(0, 3, (8,))
dataset = TensorDataset(X, y)

ARGS = {
    'model': DummyModel(device='cpu'),
    'dataset': dataset,
    'epochs': 2,
    'stop_epochs': 1,
    'lr': 0.001,
    'batch_size': 2,
    'device': 'cpu'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family 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
train Not yet run
unverified (harvested, no recorded run)
xiaofeng-github/CLUBench/CLUBench/algorithms/DEC.py
code served (permissive licence) · get_code("25747a9d381c8339")
How it ran

no invocation recorded.

train Not yet run
unverified (harvested, no recorded run)
xiaofeng-github/CLUBench/CLUBench/algorithms/IDEC.py
code served (permissive licence) · get_code("e26e1f92ac47c906")
How it ran

no invocation recorded.

Also present in other papers' code

Of 15 samples lifted from this paper's linked repositories, 15 carry a function identity and 0 do not. Of the 15 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 15 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

Clustering is a fundamental problem in data science with a long-standing research history, yielding numerous insightful algorithms. Despite this progress, a systematic and large-scale empirical evaluation that jointly considers conventional algorithms, deep learning-based methods, and recent foundation model-based clustering remains largely absent, leading to limited guidance on algorithm selection and deployment. To address this gap, we introduce CLUBench, a comprehensive clustering benchmark comprising 24 algorithms of diverse principles evaluated on 131 datasets across tabular, text, and image data, involving 178,815 experiments. Importantly, our analyses of (i) the impact of hyperparameter tuning, (ii) the impact of data types and characteristics, (iii) the impact of pretrained embeddings, (iv) large language model-based clustering, (v) the similarity of algorithms, and (vi) the low-rank structures of performance matrices, yield meaningful insights and promising pathways for clustering research. For instance, our study reveals that: 1) All evaluated deep clustering methods do not exhibit a significant advantage compared with the top-performing conventional clustering algorithms (e.g., KMeans, SpeClu) in terms of average performance; 2) For image and text clustering tasks, combining pretrained embeddings with conventional clustering algorithms (e.g., KMeans, SpeClu) offers effective and efficient clustering; 3) Clustering remains a challenging and nontrivial problem, even in the era of increasingly dominant foundation models. Moreover, we propose to use the low-rank structure in cross-model performance matrices to efficiently approximate the overall performance evaluation in practical applications. We further demonstrate the feasibility of model selection based on the performance matrices across all hyperparameter configurations. Finally, we provide an easy-to-use toolbox by encapsulating the source code from the official code repository into a unified framework, accompanied by detailed instructions. All benchmark datasets, overall experimental results, and the toolbox are fully open-sourced and available at https://github.com/xiaofeng-github/CLUBench.

For agents

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

get_harvested_code_for_paper("2605.29933")
get_code_for_paper("2605.29933")
have("2605.29933")

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

curl https://syntology.ai/api/ran/2605.29933.json

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

[![Syntology run record](https://syntology.ai/api/ran/2605.29933.svg)](https://syntology.ai/paper/2605.29933)

Connect an agent — have() is free.