SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.04810 · NeurIPS · 2022

Benchmarking Graphormer on Large-Scale Molecular Modeling Datasets

Di He, Tie-Yan Liu, Shengjie Luo, Chang Liu, Guolin Ke, Shuxin Zheng, Yu Shi, Yifei Shen, Jiacheng You, Jiyan He

arXiv · PDF · Open in the Atlas

Code that ran

8 of 9 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, 1 violated it and 3 have no contract verdict; 4 failures 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/2203.04810.json · badge: /api/ran/2203.04810.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
Microsoft/Graphormer canonical (classifier label)
classifier basis not recorded on the link
5 ran · 0 no recorded run
leffff/graphormer-pyg community (classifier label)
classifier basis not recorded on the link
3 ran · 1 no recorded run
FunctionStatusWhere it lives
axis_angle_to_matrix Ran
ran; our contract draft was wrong, not the code
Microsoft/Graphormer/distributional_graphormer/protein/model/geometry.py
code served (permissive licence) · get_code("07f28efca79efafa")
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, 3, 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)

# axis_angle_to_matrix takes a single parameter: axis_angle
# It should be a tensor of shape (..., 3) representing rotations in axis-angle form
# The magnitude of each vector is the rotation angle in radians

# Create a small batch of axis-angle vectors
# Shape: (2, 3) - batch of 2 rotation vectors, each with 3 components
axis_angle = torch.randn(2, 3)

ARGS = {
    "axis_angle": axis_angle
}
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, 3, 3) — shape (2, 3, 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
axis_angle_to_quaternion Ran
ran; our contract draft was wrong, not the code
Microsoft/Graphormer/distributional_graphormer/protein/model/geometry.py
code served (permissive licence) · get_code("e79a535a0171afe6")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned 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)

# axis_angle_to_quaternion expects:
# - axis_angle: tensor of shape (..., 3) representing rotations as axis-angle vectors

# Create a small batch of axis-angle rotations
# Shape: (2, 3) - batch of 2 rotations, each with 3 components (axis * angle)
axis_angle = torch.randn(2, 3)

ARGS = {
    "axis_angle": axis_angle,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type 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
dot_product Ran
ran on a synthesized input
leffff/graphormer-pyg/graphormer/layers.py
code served (permissive licence) · get_code("805320773bb33ccf")
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)

x1 = torch.randn(2, 4)
x2 = torch.randn(2, 4)

ARGS = {
    "x1": x1,
    "x2": x2,
}
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
floyd_warshall_source_to_all Ran
ran on a synthesized input
leffff/graphormer-pyg/graphormer/functional.py
code served (permissive licence) · get_code("5e9e26893b8f2264")
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 networkx as nx

# Create a simple directed graph for testing
G = nx.DiGraph()
G.add_edges_from([
    (0, 1),
    (1, 2),
    (0, 2),
    (2, 3)
])

source = 0
cutoff = None

ARGS = {
    'G': G,
    'source': source,
    'cutoff': cutoff
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
networkx — 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
gaussian Ran
ran, violated the contract we drafted
Microsoft/Graphormer/graphormer/models/graphormer_3d.py
pointer only (licence: MIT); the stored copy get_code serves, from bowenwang77/DR-Label, is pointer only (licence: NONE) · get_code("294ba20b052bfbc7")
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 bowenwang77/DR-Label, 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.

quaternion_to_matrix Ran
ran; our contract draft was wrong, not the code
Microsoft/Graphormer/distributional_graphormer/protein/model/geometry.py
code served (permissive licence) · get_code("757d8fbfcc15222b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned Tensor, ndim 3, dtype family float, shape (2, 3, 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 a batch of quaternions with shape (2, 4)
# Quaternions are represented as [w, x, y, z] where w is the real part
# We need to normalize them for valid rotation quaternions
quaternions = torch.randn(2, 4)
# Normalize to unit quaternions
quaternions = quaternions / torch.norm(quaternions, dim=-1, keepdim=True)

ARGS = {
    "quaternions": quaternions
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 3, dtype family float, shape (2, 3, 3) — shape (2, 3, 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
shortest_path_distance Ran
ran on a synthesized input
leffff/graphormer-pyg/graphormer/functional.py
code served (permissive licence) · get_code("dfcaec0b8cc0a3ff")
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 torch
import torch.nn as nn
from torch_geometric.data import Data
import networkx as nx

torch.manual_seed(0)

# Create a simple graph as a torch_geometric Data object
# The function converts this to networkx and computes shortest paths
edge_index = torch.tensor([
    [0, 1, 1, 2, 2, 3],
    [1, 0, 2, 1, 3, 2]
], dtype=torch.long)

node_features = torch.randn(4, 3)  # 4 nodes, 3 features
edge_features = torch.randn(6, 2)  # 6 edges, 2 features

data = Data(
    x=node_features,
    edge_index=edge_index,
    edge_attr=edge_features
)

ARGS = {
    "data": data
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
networkx, 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, 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
softmax_dropout Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
Microsoft/Graphormer/graphormer/models/graphormer_3d.py
pointer only (licence: MIT); the stored copy get_code serves, from bowenwang77/DR-Label, is pointer only (licence: NONE) · get_code("74ba23674705912e")
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 bowenwang77/DR-Label, 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.

all_pairs_shortest_path Not yet run
unverified (harvested, no recorded run)
leffff/graphormer-pyg/graphormer/functional.py
code served (permissive licence) · get_code("d6a089b5aec0d18d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 5 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 5 later) and 4 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
gaussian 12 other papers
12 later
2210.01765 (2022-10)
One Transformer Can Understand Both 2D & 3D Molecular Data
2301.09505 (2023-01)
Rethinking the Expressive Power of GNNs via Graph Biconnectivity
2303.02875 (2023-03)
DR-Label: Improving GNN Models for Catalysis Systems by Label Deconstruction and Reconstruction
and 9 more not named here
quaternion_to_matrix 6 other papers
6 later · 1 only as an assembly brick
2209.07589 (2022-09)
PIZZA: A Powerful Image-only Zero-Shot Zero-CAD Approach to 6 DoF Tracking
2210.01776 (2022-10)
DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking
2409.12140 (2024-09)
MoRAG -- Multi-Fusion Retrieval Augmented Generation for Human Motion
and 3 more not named here
axis_angle_to_matrix 3 other papers
3 later · 1 only as an assembly brick
2207.05053 (2022-07)
Learning Continuous Grasping Function with a Dexterous Hand from Human Demonstrations
2210.01776 (2022-10)
DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking
2502.02513 (2025-02)
Diffusion Generative Modeling on Lie Group Representations
softmax_dropout 3 other papers
3 later
2301.09505 (2023-01)
Rethinking the Expressive Power of GNNs via Graph Biconnectivity
2303.02875 (2023-03)
DR-Label: Improving GNN Models for Catalysis Systems by Label Deconstruction and Reconstruction
2507.00846 (2025-07)
BoltzNCE: Learning Likelihoods for Boltzmann Generation with Stochastic Interpolants and Noise Contrastive Estimation
axis_angle_to_quaternion 2 other papers
2 later · 1 only as an assembly brick
2210.01776 (2022-10)
DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking
2502.02513 (2025-02)
Diffusion Generative Modeling on Lie Group Representations

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

This technical note describes the recent updates of Graphormer, including architecture design modifications, and the adaption to 3D molecules. With these simple modifications, Graphormer attained better results on large-scale molecular modeling datasets, and the performance gain could be consistently obtained on 2D and 3D molecular graph modeling tasks. In addition, we show that with a global receptive field and an adaptive aggregation strategy, Graphormer is more powerful than classic message-passingbased GNNs. Empirically, Graphormer could achieve much less MAE than the originally reported results on the PCQM4M quantum chemistry dataset used in KDD Cup 2021. In the meanwhile, it greatly outperforms the competitors in the recent Open Catalyst Challenge, which is a competition track on NeurIPS 2021 workshop, and aims to model the catalyst-adsorbate reaction system with advanced AI models. All code can be found at this link.

For agents

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

get_harvested_code_for_paper("2203.04810")
get_code_for_paper("2203.04810")
have("2203.04810")

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

curl https://syntology.ai/api/ran/2203.04810.json

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

[![Syntology run record](https://syntology.ai/api/ran/2203.04810.svg)](https://syntology.ai/paper/2203.04810)

Connect an agent — have() is free.