SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2502.02415 · 2025

Fast Graph Generation via Autoregressive Noisy Filtration Modeling

arXiv · PDF · Open in the Atlas

Code that ran

14 of 14 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 13 have no contract verdict; 0 failures of our instrument; 0 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"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/2502.02415.json · badge: /api/ran/2502.02415.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
borgwardtlab/anfm canonical (classifier label)
classifier basis not recorded on the link
14 ran · 0 no recorded run
FunctionStatusWhere it lives
add_node_attr Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/models/robust_rwpe.py
code served (permissive licence) · get_code("ae336c57495df41b")
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 Data. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
from torch_geometric.data import Data

torch.manual_seed(0)

# Test case 1: Adding a node attribute when data.x exists (1D case)
data_1d = Data()
data_1d.x = torch.randn(2)  # 1D tensor with 2 nodes
value_1d = torch.randn(2, 1)  # Value to concatenate

# Test case 2: Adding a node attribute when data.x exists (2D case)
data_2d = Data()
data_2d.x = torch.randn(2, 4)  # 2D tensor with 2 nodes, 4 features
value_2d = torch.randn(2, 3)  # Value to concatenate

# Test case 3: Adding a named attribute
data_named = Data()
data_named.x = torch.randn(2, 4)
value_named = torch.randn(2, 5)

# Use the 1D case as the primary test
ARGS = {
    "data": data_1d,
    "value": value_1d,
    "attr_name": None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Data — a Data 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
dense_to_sparse_adj Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/utils.py
code served (permissive licence) · get_code("1096411b9848920f")
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
from torch_geometric.utils import cumsum, scatter

torch.manual_seed(0)

# Create a small dense adjacency matrix: batch_size=2, num_nodes=4
# Shape: (batch_size, num_nodes, num_nodes)
dense_adj = torch.tensor([
    [[0.0, 1.0, 0.0, 0.5],
     [1.0, 0.0, 2.0, 0.0],
     [0.0, 2.0, 0.0, 1.5],
     [0.5, 0.0, 1.5, 0.0]],
    
    [[0.0, 3.0, 0.0, 0.0],
     [3.0, 0.0, 0.5, 0.0],
     [0.0, 0.5, 0.0, 2.0],
     [0.0, 0.0, 2.0, 0.0]]
], dtype=torch.float32)

# Create a mask: shape (batch_size, num_nodes) to selectively include nodes
mask = torch.tensor([
    [True, True, True, False],
    [True, True, False, True]
], dtype=torch.bool)

ARGS = {
    "dense_adj": dense_adj,
    "mask": mask,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — 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
dense_to_sparse_batch Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/utils.py
code served (permissive licence) · get_code("dd3521a053c14bb8")
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 (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
from torch_geometric.utils import cumsum, scatter

torch.manual_seed(0)

# Create a dense batch tensor: bs=2, num_nodes=4, num_features=3
x = torch.randn(2, 4, 3)

# Create a mask tensor: bs=2, num_nodes=4
# First batch has 3 valid nodes, second batch has 2 valid nodes
mask = torch.tensor([
    [True, True, True, False],
    [True, True, False, False]
], dtype=torch.bool)

ARGS = {
    "x": x,
    "mask": mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — 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 (5, 3) — shape (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
entropy_lower_bound Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/models/edge_predictor.py
code served (permissive licence) · get_code("9bba061985b984f2")
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 float, shape (2,) 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)

# Define small dimensions for each semantic dimension
t = 2  # time steps
num_components = 3
num_nodes = 4
num_edge_types = 5
# star_shape will be empty for simplicity

# Construct tensors with the required shapes
# mixture_logits: (t, num_components)
mixture_logits = torch.randn(t, num_components)

# logits: (t, num_components, num_nodes, num_nodes, num_edge_types)
logits = torch.randn(t, num_components, num_nodes, num_nodes, num_edge_types)

# mask: (t, num_nodes)
mask = torch.randint(0, 2, (t, num_nodes), dtype=torch.bool)

ARGS = {
    "mixture_logits": mixture_logits,
    "logits": logits,
    "mask": mask,
}
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 float, shape (2,) — shape (2,) 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
entropy_upper_bound Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/models/edge_predictor.py
code served (permissive licence) · get_code("115ae97400fcb695")
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 float, shape (2,) 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)

# Define small tensor dimensions
t = 2  # time steps
num_components = 3
num_nodes = 4
num_edge_types = 5

# No star_shape (empty), so shapes are:
# mixture_logits: (t, num_components)
# logits: (t, num_components, num_nodes, num_nodes, num_edge_types)
# mask: (t, num_nodes)

mixture_logits = torch.randn(t, num_components)
logits = torch.randn(t, num_components, num_nodes, num_nodes, num_edge_types)
mask = torch.randint(0, 2, (t, num_nodes)).bool()

ARGS = {
    "mixture_logits": mixture_logits,
    "logits": logits,
    "mask": mask,
}
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 float, shape (2,) — shape (2,) 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
evaluate_discriminator Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/train_gan.py
code served (permissive licence) · get_code("d766e30cf076eca4")
How it ran

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

import torch
import torch.nn as nn
import numpy as np

torch.manual_seed(0)

class SimpleLightningFabric:
    """Stand-in for a Lightning Fabric object."""
    def all_reduce(self, metrics, reduce_op='mean'):
        """Return metrics as-is for single-GPU simulation."""
        return metrics

class SimpleDiscriminator(nn.Module):
    """Minimal discriminator that outputs logits."""
    def __init__(self, input_dim=4):
        super().__init__()
        self.linear = nn.Linear(input_dim, 1)
    
    def forward(self, x):
        return self.linear(x)

fabric = SimpleLightningFabric()

discriminator = SimpleDiscriminator(input_dim=4)

positive_batch = torch.randn(2, 4)

negative_batch = torch.randn(2, 4)

ARGS = {
    'fabric': fabric,
    'discriminator': discriminator,
    'positive_batch': positive_batch,
    'negative_batch': negative_batch,
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_gradient_magnitudes Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/train_gan.py
code served (permissive licence) · get_code("cef551adde62737a")
How it ran

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

import torch
import torch.nn as nn
import numpy as np

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

# Create a model and perform a forward/backward pass to populate gradients
model = SimpleModel()
dummy_input = torch.randn(2, 4)
dummy_target = torch.randn(2, 2)

output = model(dummy_input)
loss = ((output - dummy_target) ** 2).mean()
loss.backward()

ARGS = {
    'model': model,
    'quantile_levels': (0.01, 0.1, 0.5, 0.9, 0.99, 0.999)
}
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 dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
get_model Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/models/sat.py
code served (permissive licence) · get_code("e86d25cc1327aca6")
How it ran

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

import torch
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal nn.Module subclass that can be called as a class
class SimpleModel(nn.Module):
    def __init__(self, num_layers, embed_dim, **kwargs):
        super().__init__()
        self.num_layers = num_layers
        self.embed_dim = embed_dim
        self.linear = nn.Linear(embed_dim, embed_dim)
    
    def forward(self, x):
        return self.linear(x)

ARGS = {
    'cls': SimpleModel,
    'name': 'small'
}
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 SimpleModel — a SimpleModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleModel output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
l2 Ran
ran, honoured the contract we drafted
borgwardtlab/anfm/anfm/evaluation/dist_helper.py
pointer only (licence: MIT); the stored copy get_code serves, from lf-wen/hgdm, is pointer only (licence: NONE) · get_code("039fc6561a9516bc")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (MIT), but get_code serves the stored copy, first harvested from lf-wen/hgdm, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

labeled_graph_collate Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/train_gan.py
code served (permissive licence) · get_code("afd65063d68582cc")
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
from torch_geometric.data import Data, Batch

torch.manual_seed(0)

# Create sample graph data
# Each sample is a tuple of (Data object, label)
data1 = Data(
    x=torch.randn(4, 3),  # 4 nodes, 3 features
    edge_index=torch.tensor([[0, 1, 2], [1, 2, 3]], dtype=torch.long)
)
label1 = 1.0

data2 = Data(
    x=torch.randn(5, 3),  # 5 nodes, 3 features
    edge_index=torch.tensor([[0, 1, 2, 3], [1, 2, 3, 4]], dtype=torch.long)
)
label2 = 0.0

samples = [(data1, label1), (data2, label2)]

ARGS = {
    "samples": samples
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — 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
modulate Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/models/sat.py
code served (permissive licence) · get_code("d2ca4f87058cba1d")
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, 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

torch.manual_seed(0)

# Case 1: x is 3D (batch, seq, dim)
x_3d = torch.randn(2, 5, 4)  # batch=2, seq=5, dim=4
shift_3d = torch.randn(2, 4)  # batch=2, dim=4
scale_3d = torch.randn(2, 4)  # batch=2, dim=4

# Case 2: x is 4D (batch, channels, height, width)
x_4d = torch.randn(2, 4, 3, 3)  # batch=2, channels=4, height=3, width=3
shift_4d = torch.randn(2, 4)  # batch=2, channels=4
scale_4d = torch.randn(2, 4)  # batch=2, channels=4

# Using 3D case for the main ARGS
ARGS = {
    "x": x_3d,
    "shift": shift_3d,
    "scale": scale_3d,
}
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, 4) — shape (2, 5, 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
sample_edges Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/models/edge_predictor.py
code served (permissive licence) · get_code("6bc1ea52fba8e7a7")
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 int, shape (2, 4, 4, 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 torch.distributions as D

torch.manual_seed(0)

# Set up dimensions
batch_size = 2
num_components = 3
num_nodes = 4
num_edge_types = 5

# logits: shape (batch_size, num_components, num_nodes, num_nodes, num_edge_types)
logits = torch.randn(batch_size, num_components, num_nodes, num_nodes, num_edge_types)

# mixture_logits: shape (batch_size, num_components)
mixture_logits = torch.randn(batch_size, num_components)

# mask: shape (batch_size, num_nodes)
# Boolean mask indicating which nodes are valid
mask = torch.ones(batch_size, num_nodes, dtype=torch.bool)

ARGS = {
    "logits": logits,
    "mixture_logits": mixture_logits,
    "mask": mask,
}
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 int, shape (2, 4, 4, 5) — shape (2, 4, 4, 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
sparse_to_dense_adj Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/utils.py
code served (permissive licence) · get_code("577a3f476486c5e8")
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
from torch_geometric.utils import cumsum, scatter

torch.manual_seed(0)

# Create a simple example with 2 graphs in a batch
# Graph 0: 3 nodes, Graph 1: 2 nodes

# edge_index: shape [2, num_edges] - edges for both graphs combined
# First 2 edges belong to graph 0, next 2 edges belong to graph 1
edge_index = torch.tensor([
    [0, 1, 3, 4],  # source nodes (with global indexing)
    [1, 2, 4, 3]   # target nodes (with global indexing)
], dtype=torch.long)

# batch: shape [total_nodes] - indicates which graph each node belongs to
# Nodes 0,1,2 belong to graph 0; nodes 3,4 belong to graph 1
batch = torch.tensor([0, 0, 0, 1, 1], dtype=torch.long)

# max_num_nodes: maximum number of nodes in any graph
max_num_nodes = 3

# edge_attr: optional edge attributes, shape [num_edges, feature_dim]
edge_attr = torch.ones(4, 2, dtype=torch.float32)

ARGS = {
    "edge_index": edge_index,
    "batch": batch,
    "max_num_nodes": max_num_nodes,
    "edge_attr": edge_attr
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, torch_geometric — 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
to_sparse_batch Ran
ran on a synthesized input
borgwardtlab/anfm/anfm/models/mixer_model.py
code served (permissive licence) · get_code("722ba5d89ebf60e0")
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
from einops import rearrange

torch.manual_seed(0)

# Construct arguments for to_sparse_batch
# The function takes:
# - x: a tensor that will be rearranged from 'b n d' to '(b n) d'
# - mask: a boolean mask that will be flattened and used to index x

# Create a small tensor with shape (b=2, n=3, d=4)
# This represents a batch of 2, with 3 elements per batch, and 4 dimensions
x = torch.randn(2, 3, 4)

# Create a mask with shape (b=2, n=3) that will select some elements
# Let's make it so some elements are True and some are False
mask = torch.tensor([
    [True, False, True],
    [False, True, True]
])

ARGS = {
    "x": x,
    "mask": mask
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
einops, 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

Also present in other papers' code

Of 14 samples lifted from this paper's linked repositories, 14 carry a function identity and 0 do not. Of the 14 distinct functions with an identity, 1 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 1 later) and 13 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
l2 10 other papers
9 dated earlier, 1 later
1910.00760 (2019-10)
Efficient Graph Generation with Graph Recurrent Attention Networks
2204.01613 (2022-04)
SPECTRE: Spectral Conditioning Helps to Overcome the Expressivity Limits of One-shot Graph Generators
2305.19337 (2023-05)
HiGen: Hierarchical Graph Generative Networks
and 7 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

Existing graph generative models often face a critical trade-off between sample quality and generation speed. We introduce Autoregressive Noisy Filtration Modeling (ANFM), a flexible autoregressive framework that addresses both challenges. ANFM leverages filtration, a concept from topological data analysis, to transform graphs into short sequences of subgraphs. We identify exposure bias as a potential hurdle in autoregressive graph generation and propose noise augmentation and reinforcement learning as effective mitigation strategies, which allow ANFM to learn both edge addition and deletion operations. This unique capability enables ANFM to correct errors during generation by modeling non-monotonic graph sequences. Our results show that ANFM matches state-of-the-art diffusion models in quality while offering over 100 times faster inference, making it a promising approach for high-throughput graph generation. The source code is publicly available at https://github.com/BorgwardtLab/anfm .

For agents

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

get_harvested_code_for_paper("2502.02415")
get_code_for_paper("2502.02415")
have("2502.02415")

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

curl https://syntology.ai/api/ran/2502.02415.json

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

[![Syntology run record](https://syntology.ai/api/ran/2502.02415.svg)](https://syntology.ai/paper/2502.02415)

Connect an agent — have() is free.