SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2111.12840 · NeurIPS · 2021

Explaining machine-learned particle-flow reconstruction

Jean-Roch Vlimant, Joosep Pata, Raghav Kansal, Javier Duarte, Maurizio Pierini, Farouk Mokhtar, Daniel Diaz

arXiv · PDF · Open in the Atlas

Code that ran

17 of 18 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 16 have no contract verdict; 1 failure of our instrument; 1 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/2111.12840.json · badge: /api/ran/2111.12840.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
faroukmokhtar/particleflow canonical (classifier label)
classifier basis not recorded on the link
17 ran · 1 no recorded run
FunctionStatusWhere it lives
classification_loss Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/losses.py
code served (permissive licence) · get_code("e64cf35ab8bff4bb")
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.functional as F

torch.manual_seed(0)

# Construct the y dictionary with cls_id tensor
# cls_id should be a 1D tensor of class labels (batch_size,)
# Using batch size 2, with class IDs ranging from 0 to 3
y = {
    'cls_id': torch.tensor([0, 1, 2, 0], dtype=torch.long)  # batch_size=4
}

# Construct the ypred dictionary with required tensors
# ypred['cls_binary'] should be logits for binary classification (batch_size, 2)
# ypred['cls_id_onehot'] should be logits for multi-class (batch_size, num_classes)
num_classes = 4
batch_size = 4

ypred = {
    'cls_binary': torch.randn(batch_size, 2),  # binary classification logits
    'cls_id_onehot': torch.randn(batch_size, num_classes)  # multi-class logits
}

ARGS = {
    'y': y,
    'ypred': ypred
}
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 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
customize_pipeline_test Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/customizations.py
code served (permissive licence) · get_code("2b270dba64c8bbb1")
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 numpy as np

torch.manual_seed(0)

# Build a minimal config dict that satisfies the function's requirements
config = {
    'batching': {
        'bucket_by_sequence_length': True,
    },
    'datasets': ['cms_pf_ttbar'],
    'train_test_datasets': {
        'physical': {
            'datasets': ['cms_pf_ttbar'],
            'batch_per_gpu': 4,
        }
    },
    'validation_dataset': 'cms_pf_ttbar',
    'validation_batch_size': 2,
    'evaluation_datasets': {
        'cms_pf_ttbar': {
            'batch_size': 2,
            'num_events': -1,
        }
    },
    'parameters': {
        'num_graph_layers_id': 5,
        'num_graph_layers_cls': 3,
    },
}

ARGS = {
    'config': config,
}
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
dataset_input_type_id Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/conf.py
code served (permissive licence) · get_code("c01f983334e0f6b8")
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 int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

# Construct arguments for dataset_input_type_id
# The function takes a dataset_name string parameter and returns an int

ARGS = {
    'dataset_name': 'test_hits_dataset'
}
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 int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
dataset_source_id Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/conf.py
code served (permissive licence) · get_code("d8a81f0febeff149")
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 int, ndim 0, dtype family int, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

# Setup code for dataset_source_id function

from enum import Enum

class Dataset(Enum):
    CMS = 'cms'
    CLIC = 'clic'
    CLD = 'cld'
    CLIC_HITS = 'clic_hits'
    CLD_HITS = 'cld_hits'

SOURCE_IDS = {'unknown': 0, Dataset.CMS.value: 1, Dataset.CLIC.value: 2, Dataset.CLD.value: 3}

def dataset_source_id(dataset_name: str) -> int:
    for source_name, source_id in SOURCE_IDS.items():
        if source_name != 'unknown' and dataset_name.startswith(f'{source_name}_'):
            return source_id
    return SOURCE_IDS['unknown']

# Test arguments
ARGS = {
    'dataset_name': 'cms_data'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type int, ndim 0, dtype family int, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
deltaphi Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/jet_utils.py
code served (permissive licence) · get_code("72d34ad7b15b8a79")
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 float, 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 numpy as np
import numba

@numba.njit
def deltaphi(phi1, phi2):
    diff = phi1 - phi2
    return np.arctan2(np.sin(diff), np.cos(diff))

# The function takes two angle arguments (phi1 and phi2) and computes the difference
# wrapped to [-pi, pi] using arctan2. Both should be numpy arrays or scalars.

# Create small test arrays representing angles in radians
phi1_val = np.array([0.5, 1.0, 1.5, 2.0], dtype=np.float64)
phi2_val = np.array([0.2, 0.8, 1.2, 1.8], dtype=np.float64)

ARGS = {
    "phi1": phi1_val,
    "phi2": phi2_val,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, numpy — 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 float, 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
deltar Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/jet_utils.py
code served (permissive licence) · get_code("48d587451f46aabd")
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 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 numpy as np
import numba

@numba.njit
def deltaphi(phi1, phi2):
    diff = phi1 - phi2
    return np.arctan2(np.sin(diff), np.cos(diff))

@numba.njit
def deltar(eta1, phi1, eta2, phi2):
    deta = eta1 - eta2
    dphi = deltaphi(phi1, phi2)
    return np.sqrt(deta ** 2 + dphi ** 2)

# Create small numpy arrays for the parameters
eta1 = np.array([1.5, 2.3], dtype=np.float64)
phi1 = np.array([0.5, 1.2], dtype=np.float64)
eta2 = np.array([1.0, 2.0], dtype=np.float64)
phi2 = np.array([0.3, 1.0], dtype=np.float64)

ARGS = {
    "eta1": eta1,
    "phi1": phi1,
    "eta2": eta2,
    "phi2": phi2
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, numpy — 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 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
get_regions Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/heptv2.py
code served (permissive licence) · get_code("9c993588b9c2e6fc")
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, 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
from einops import rearrange

torch.manual_seed(0)

ARGS = {
    "num_regions": 8,
    "num_or_hashes": 2,
    "num_heads": 3,
    "num_and_hashes": 2,
}
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 3, dtype family float, shape (2, 2, 3) — shape (2, 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
invert_permutation Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/heptv2.py
code served (permissive licence) · get_code("ded74e6af3ddb339")
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 int, 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
import os

torch.manual_seed(0)

# Create a small permutation tensor
# A permutation is a rearrangement of indices
# Shape: (2, 4) - batch size 2, dimension 4
perm = torch.tensor([[1, 3, 0, 2], [2, 0, 3, 1]], dtype=torch.long)

ARGS = {
    'perm': perm,
}
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 int, 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
lsh_mapping Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/hept.py
code served (permissive licence) · get_code("a99434c75350faa3")
How it ran

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

import torch
import torch.nn as nn

torch.manual_seed(0)

class SimpleHasher(nn.Module):
    """A minimal LSH hasher that maps input tensors to hash values."""
    def __init__(self, input_dim, output_dim):
        super().__init__()
        self.projection = nn.Linear(input_dim, output_dim)
    
    def forward(self, x):
        # Apply projection and return hash-like values
        return self.projection(x)

# Create the e2lsh model
input_dim = 4
output_dim = 3
e2lsh = SimpleHasher(input_dim, output_dim)

# Create queries and keys tensors
# queries: shape (2, 4) - batch of 2, dimension 4
# keys: shape (2, 4) - batch of 2, dimension 4
queries = torch.randn(2, 4)
keys = torch.randn(2, 4)

ARGS = {
    "e2lsh": e2lsh,
    "queries": queries,
    "keys": keys,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
match_jets Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/jet_utils.py
code served (permissive licence) · get_code("45ca4d36fd5a0792")
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 numpy as np

class Jet:
    def __init__(self, eta, phi):
        self.eta = eta
        self.phi = phi

# Create two events with jets
# Event 1: 3 jets in first collection, 2 jets in second collection
jets1_event1 = Jet(
    eta=np.array([0.5, 1.2, -0.8]),
    phi=np.array([0.1, 1.5, 2.3])
)
jets2_event1 = Jet(
    eta=np.array([0.55, 2.0]),
    phi=np.array([0.12, 1.6])
)

# Event 2: 2 jets in first collection, 3 jets in second collection
jets1_event2 = Jet(
    eta=np.array([1.0, -1.5]),
    phi=np.array([0.5, 2.8])
)
jets2_event2 = Jet(
    eta=np.array([1.05, -1.48, 0.2]),
    phi=np.array([0.51, 2.79, 1.2])
)

jets1 = [jets1_event1, jets1_event2]
jets2 = [jets2_event1, jets2_event2]

deltaR_cut = 0.5

ARGS = {
    "jets1": jets1,
    "jets2": jets2,
    "deltaR_cut": deltaR_cut,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numba, numpy — 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
pairwise_l2_dist Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/gnnlsh.py
code served (permissive licence) · get_code("c22d7ce353754a1e")
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

torch.manual_seed(0)

# Create two small tensors for pairwise L2 distance computation
# A: shape (2, 4) - 2 points in 4D space
# B: shape (3, 4) - 3 points in 4D space
A = torch.randn(2, 4, dtype=torch.float32)
B = torch.randn(3, 4, dtype=torch.float32)

ARGS = {
    "A": A,
    "B": B,
}
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
point_wise_feed_forward_network Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/gnnlsh.py
code served (permissive licence) · get_code("a3aa89db788edd6f")
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 Sequential. 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 import nn

torch.manual_seed(0)

ARGS = {
    "d_in": 4,
    "d_hidden": 8,
    "d_out": 3,
    "num_layers": 2,
    "activation": "ELU",
    "dropout": 0.1
}
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 Sequential — a Sequential output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a Sequential output the recorded ndim 1 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
regression_loss Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/losses.py
code served (permissive licence) · get_code("5c5bebafe8dccc63")
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.functional as F

torch.manual_seed(0)

# Construct y dictionary with required keys
# y['cls_id']: classification ID (0 means not a particle)
# y[feature]: target values for each regression feature
batch_size = 2
y = {
    'cls_id': torch.tensor([1, 0, 1, 1]),  # Mix of particles (1) and non-particles (0)
    'pt': torch.randn(4),
    'eta': torch.randn(4),
    'sin_phi': torch.randn(4),
    'cos_phi': torch.randn(4),
    'energy': torch.randn(4),
}

# Construct ypred dictionary with predictions for each feature
ypred = {
    'pt': torch.randn(4),
    'eta': torch.randn(4),
    'sin_phi': torch.randn(4),
    'cos_phi': torch.randn(4),
    'energy': torch.randn(4),
}

# input_pt: base transverse momentum for scaling
input_pt = torch.tensor([50.0, 75.0, 100.0, 120.0])

# regression_weights: weight multiplier for each feature
regression_weights = {
    'pt': 1.0,
    'eta': 0.5,
    'sin_phi': 0.3,
    'cos_phi': 0.3,
    'energy': 1.0,
}

ARGS = {
    'y': y,
    'ypred': ypred,
    'input_pt': input_pt,
    'regression_weights': regression_weights,
}
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 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
sliced_wasserstein_loss Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/losses.py
code served (permissive licence) · get_code("cfb51c77ac54ffd0")
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

torch.manual_seed(0)

# Construct tiny tensors for y_pred and y_true
# Shape: (batch_size, feature_dim)
# Using batch_size=2, feature_dim=4
batch_size = 2
feature_dim = 4

y_true = torch.randn(batch_size, feature_dim)
y_pred = torch.randn(batch_size, feature_dim)

num_projections = 200

ARGS = {
    "y_pred": y_pred,
    "y_true": y_true,
    "num_projections": num_projections,
}
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
split_indices_to_bins_batch Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/gnnlsh.py
code served (permissive licence) · get_code("b146c0426554f03c")
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 int, shape (2, 16, 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)

batch_size = 2
n_points = 8
num_or_hashes = 4
bin_size = 2

# Create tensor 'a' with shape (batch_size, n_points, num_or_hashes)
a = torch.randn(batch_size, n_points, num_or_hashes)

# Create mask 'msk' with shape (batch_size, n_points)
# Should have some True and False values
msk = torch.randint(0, 2, (batch_size, n_points)).to(torch.bool)

ARGS = {
    'a': a,
    'bin_size': bin_size,
    'msk': msk,
    'stable_sort': 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 Tensor, ndim 3, dtype family int, shape (2, 16, 2) — shape (2, 16, 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
trunc_normal_ Ran
ran on a synthesized input
faroukmokhtar/particleflow/mlpf/model/hept.py
code served (permissive licence) · get_code("e932b7d3ef0f378b")
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; no output shape is stated. It ran in syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1) at 2026-09-26T22:17:14Z. 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 math
import torch

torch.manual_seed(0)

# Create a small tensor to be filled with truncated normal values
tensor = torch.empty(2, 4)

# Define arguments for trunc_normal_
ARGS = {
    "tensor": tensor,
    "mean": 0.0,
    "std": 1.0,
    "a": -2.0,
    "b": 2.0,
}
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 Tensor, ndim 2, dtype family float — no output shape is stated: no later run recorded an output shape for this sample (observed_output and metamorphic_properties carry none; shape_contract and output_contract record ndim only)
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 syntology-harvest-sandbox:v5-deps (sha256:6a292aa1aed1); 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
uniform Ran
ran; our contract draft was wrong, not the code
faroukmokhtar/particleflow/mlpf/model/hept.py
code served (permissive licence) · get_code("1d809fe0159efa28")
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)

ARGS = {
    'a': 0.0,
    'b': 10.0,
    'shape': (2, 4),
    'device': 'cpu'
}
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
quantile_partition Not yet run
unverified (harvested, no recorded run)
faroukmokhtar/particleflow/mlpf/model/heptv2.py
code served (permissive licence) · get_code("574204301da463f9")
How it ran

no invocation recorded.

Also present in other papers' code

Of 18 samples lifted from this paper's linked repositories, 18 carry a function identity and 0 do not. Of the 18 distinct functions with an identity, 1 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 1 later) and 17 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
uniform 1 other paper
1 later
2402.12535 (2024-02)
Locality-Sensitive Hashing-Based Efficient Point Transformer with Applications in High-Energy Physics

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 particle-flow (PF) algorithm is used in general-purpose particle detectors to reconstruct a comprehensive particle-level view of the collision by combining information from different subdetectors. A graph neural network (GNN) model, known as the machine-learned particle-flow (MLPF) algorithm, has been developed to substitute the rule-based PF algorithm. However, understanding the model's decision making is not straightforward, especially given the complexity of the setto-set prediction task, dynamic graph building, and message-passing steps. In this paper, we adapt the layerwise-relevance propagation technique for GNNs and apply it to the MLPF algorithm to gauge the relevant nodes and features for its predictions. Through this process, we gain insight into the model's decision-making.

For agents

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

get_harvested_code_for_paper("2111.12840")
get_code_for_paper("2111.12840")
have("2111.12840")

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

curl https://syntology.ai/api/ran/2111.12840.json

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

[![Syntology run record](https://syntology.ai/api/ran/2111.12840.svg)](https://syntology.ai/paper/2111.12840)

Connect an agent — have() is free.