SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2410.06262 · ICLR · 2025

SymDiff: Equivariant Diffusion via Stochastic Symmetrisation

Rob Cornish, Yee Teh, Leo Zhang, Kianoosh Ashouritaklimi

arXiv · PDF · Open in the Atlas

Code that ran

9 of 12 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, 1 violated it and 0 have no contract verdict; 7 failures of our instrument; 3 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/2410.06262.json · badge: /api/ran/2410.06262.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
leozhangML/SymDiff canonical (classifier label)
classifier basis: census
9 ran · 3 no recorded run
FunctionStatusWhere it lives
compute_loss_and_nll Ran
ran; our contract draft was wrong, not the code
leozhangML/SymDiff/qm9/losses.py
code served (permissive licence) · get_code("4ed06d5114046b7e")
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
from torch.distributions import Poisson

torch.manual_seed(0)

# Create a minimal args object
class Args:
    def __init__(self):
        self.probabilistic_model = 'diffusion'

args = Args()

# Create a minimal generative model that accepts the expected inputs
class GenerativeModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def forward(self, x, h, node_mask, edge_mask, context):
        # x: (bs, n_nodes, n_dims)
        # Returns nll of shape (bs,)
        bs = x.size(0)
        return torch.randn(bs)

generative_model = GenerativeModel()

# Create nodes_dist with log_prob method
class NodesDist:
    def log_prob(self, N):
        # N is a long tensor of shape (bs,)
        return torch.randn(N.size(0))

nodes_dist = NodesDist()

# Create tensors with distinct sizes
bs = 2
n_nodes = 3
n_dims = 4

x = torch.randn(bs, n_nodes, n_dims)
h = torch.randn(bs, n_nodes, 5)
node_mask = torch.ones(bs, n_nodes, 1)
edge_mask = torch.ones(bs, n_nodes, n_nodes)
context = torch.randn(bs, 6)

ARGS = {
    'args': args,
    'generative_model': generative_model,
    'nodes_dist': nodes_dist,
    'x': x,
    'h': h,
    'node_mask': node_mask,
    'edge_mask': edge_mask,
    'context': context
}
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
compute_n_nodes_dict Ran
ran; our contract draft was wrong, not the code
leozhangML/SymDiff/analyse_geom.py
code served (permissive licence) · get_code("1b43a8d3512fffdf")
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 numpy as np

# Create a small synthetic .npy file data
# The function expects data where:
# - Column 0: molecule ID (int)
# - Column 1: atom type (float, where 1.0 is hydrogen)
# - Other columns: not used in this function

# Create synthetic molecular data
# Format: [mol_id, atom_type, ...]
synthetic_data = np.array([
    [0, 6.0, 0.1, 0.2],  # mol 0, carbon
    [0, 6.0, 0.3, 0.4],  # mol 0, carbon
    [0, 1.0, 0.5, 0.6],  # mol 0, hydrogen (will be removed)
    [1, 6.0, 0.7, 0.8],  # mol 1, carbon
    [1, 8.0, 0.9, 1.0],  # mol 1, oxygen
    [1, 1.0, 1.1, 1.2],  # mol 1, hydrogen (will be removed)
    [2, 6.0, 1.3, 1.4],  # mol 2, carbon
    [2, 6.0, 1.5, 1.6],  # mol 2, carbon
    [2, 6.0, 1.7, 1.8],  # mol 2, carbon
], dtype=np.float32)

# Save to a temporary location that the function can load
import tempfile
import os

temp_dir = tempfile.gettempdir()
temp_file = os.path.join(temp_dir, 'test_geom_data.npy')
np.save(temp_file, synthetic_data)

ARGS = {
    'file': temp_file,
    'remove_hydrogens': True
}
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 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
convert_str_to_bool Ran
ran, violated the contract we drafted
leozhangML/SymDiff/eval_analyze.py
code served (permissive licence) · get_code("12f8e6ced3e7acc4")
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 bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn

torch.manual_seed(0)

ARGS = {
    'string': 'True'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, 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
get_optim Ran
ran; our contract draft was wrong, not the code
leozhangML/SymDiff/qm9/models.py
code served (permissive licence) · get_code("85c8c5cc8774ca97")
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 AdamW. 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 generative model that has .parameters() method
class MinimalGenerativeModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

# Create an instance of the generative model
generative_model = MinimalGenerativeModel()

# Create an args object with required attributes
class Args:
    def __init__(self):
        self.lr = 0.001
        self.use_amsgrad = True
        self.weight_decay = 0.01

args = Args()

ARGS = {
    "args": args,
    "generative_model": generative_model
}
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 AdamW — a AdamW output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a AdamW 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_scheduler Ran
ran; our contract draft was wrong, not the code
leozhangML/SymDiff/qm9/models.py
code served (permissive licence) · get_code("0f0e2214104b64a1")
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 LambdaLR. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
from transformers import get_cosine_schedule_with_warmup
from torch.optim import Adam

torch.manual_seed(0)

# Create a minimal model for the optimizer
model = nn.Linear(4, 3)

# Create an optimizer
optim = Adam(model.parameters(), lr=0.001)

# Create args object with required attributes
class Args:
    def __init__(self):
        self.scheduler = 'cosine'
        self.num_warmup_steps = 100
        self.num_training_steps = 1000

args = Args()

ARGS = {
    'args': args,
    'optim': optim
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, transformers — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type LambdaLR — a LambdaLR output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a LambdaLR 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
gradient_clipping Ran
ran, honoured the contract we drafted
leozhangML/SymDiff/utils.py
code served (permissive licence) · get_code("abeabf4526a83aaa")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn


class SimpleFlow(nn.Module):
    """Minimal flow model with parameters that can have gradients."""
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
        self.linear2 = nn.Linear(3, 2)
    
    def forward(self, x):
        return self.linear2(torch.relu(self.linear(x)))


class GradNormQueue:
    """Simple queue for storing gradient norms."""
    def __init__(self, maxlen=100):
        self.maxlen = maxlen
        self.queue = []
    
    def add(self, value):
        self.queue.append(value)
        if len(self.queue) > self.maxlen:
            self.queue.pop(0)
    
    def mean(self):
        if len(self.queue) == 0:
            return torch.tensor(1.0)
        return torch.tensor(sum(self.queue) / len(self.queue))
    
    def std(self):
        if len(self.queue) == 0:
            return torch.tensor(0.0)
        mean_val = self.mean()
        variance = sum((x - mean_val) ** 2 for x in self.queue) / max(1, len(self.queue))
        return torch.sqrt(variance)


torch.manual_seed(0)

# Create a simple flow model
flow = SimpleFlow()

# Create some gradients by doing a forward/backward pass
x = torch.randn(2, 4)
y = flow(x)
loss = y.sum()
loss.backward()

# Create gradient norm queue with some initial values
gradnorm_queue = GradNormQueue(maxlen=100)
for _ in range(5):
    gradnorm_queue.add(0.5)

# Create args object with max_grad_norm attribute
class Args:
    def __init__(self):
        self.max_grad_norm = 2.0

args = Args()

ARGS = {
    'args': args,
    'flow': flow,
    'gradnorm_queue': gradnorm_queue,
    'clipping_type': 'queue'
}
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
load_split_data Ran
ran; our contract draft was wrong, not the code
leozhangML/SymDiff/build_geom_dataset.py
code served (permissive licence) · get_code("3858d40af29fc810")
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 os
import numpy as np
import tempfile
from pathlib import Path

# Create temporary directory and test files
temp_dir = tempfile.mkdtemp()

# Create a small conformer data file
# Format: [mol_id, conformer_coords...]
# We'll create data with 3 molecules, each with multiple conformers
mol_data = []

# Molecule 0: 2 conformers with 4 coordinate values each
mol_data.append([0, 1.0, 2.0, 3.0, 4.0])
mol_data.append([0, 1.1, 2.1, 3.1, 4.1])

# Molecule 1: 3 conformers
mol_data.append([1, 5.0, 6.0, 7.0, 8.0])
mol_data.append([1, 5.1, 6.1, 7.1, 8.1])
mol_data.append([1, 5.2, 6.2, 7.2, 8.2])

# Molecule 2: 2 conformers
mol_data.append([2, 9.0, 10.0, 11.0, 12.0])
mol_data.append([2, 9.1, 10.1, 11.1, 12.1])

all_data = np.array(mol_data, dtype=np.float32)
conformation_file = os.path.join(temp_dir, 'conformations.npy')
np.save(conformation_file, all_data)

# Create the permutation file that the function expects
# Simple identity permutation for 3 molecules
perm = np.array([0, 1, 2], dtype=np.int32)
perm_file = os.path.join(temp_dir, 'geom_permutation.npy')
np.save(perm_file, perm)

ARGS = {
    'conformation_file': conformation_file,
    'val_proportion': 0.1,
    'test_proportion': 0.1,
    'filter_size': None
}
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 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
random_rotation Ran
ran; our contract draft was wrong, not the code
leozhangML/SymDiff/utils.py
code served (permissive licence) · get_code("e372872a463628a7")
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, 4, 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 numpy as np

torch.manual_seed(0)

# Test case 1: 2D rotation
# x should have shape (batch_size, n_nodes, n_dims)
# For 2D case: (bs, n_nodes, 2)
x_2d = torch.randn(2, 4, 2)

# Test case 2: 3D rotation
# For 3D case: (bs, n_nodes, 3)
x_3d = torch.randn(2, 4, 3)

# The function takes one parameter: x
# We'll test with the 3D case as it exercises more code
ARGS = {
    "x": x_3d
}
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 Tensor, ndim 3, dtype family float, shape (2, 4, 3) — shape (2, 4, 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
sum_except_batch Ran
ran; our contract draft was wrong, not the code
leozhangML/SymDiff/qm9/losses.py
code served (permissive licence) · get_code("90c529ca8e85d3db")
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)

# The function takes a tensor x, reshapes it to (batch_size, -1),
# and sums across the flattened dimension. We need a multi-dimensional tensor.
x = torch.randn(2, 4, 3)  # batch=2, other dims flattened to 12

ARGS = {
    "x": x
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Tensor, ndim 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_args_gen Not yet run
unverified (harvested, no recorded run)
leozhangML/SymDiff/eval_conditional_qm9.py
code served (permissive licence) · get_code("f576c0a0e0ae29c2")
How it ran

no invocation recorded.

get_dataset_info Not yet run
unverified (harvested, no recorded run)
leozhangML/SymDiff/configs/datasets_config.py
code served (permissive licence) · get_code("476ddad204b6b1b3")
How it ran

no invocation recorded.

load_model Not yet run
unverified (harvested, no recorded run)
leozhangML/SymDiff/utils.py
code served (permissive licence) · get_code("fc4da23dd75dcb46")
How it ran

no invocation recorded.

Also present in other papers' code

Of 12 samples lifted from this paper's linked repositories, 12 carry a function identity and 0 do not. Of the 12 distinct functions with an identity, 7 are also present in other papers whose code we hold (7 with at least one paper dated earlier, 3 the same month, 4 later) and 5 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
sum_except_batch 6 other papers
4 dated earlier, 1 the same month, 1 later
2105.09016 (2021-05)
E(n) Equivariant Normalizing Flows
2203.17003 (2022-03)
Equivariant Diffusion for Molecule Generation in 3D
2305.01140 (2023-05)
Geometric Latent Diffusion Models for 3D Molecule Generation
and 3 more not named here
compute_loss_and_nll 4 other papers
2 dated earlier, 1 the same month, 1 later
2203.17003 (2022-03)
Equivariant Diffusion for Molecule Generation in 3D
2305.01140 (2023-05)
Geometric Latent Diffusion Models for 3D Molecule Generation
2410.06502 (2024-10)
Chemistry-Inspired Diffusion with Non-Differentiable Guidance
and 1 more not named here
get_args_gen 4 other papers
2 dated earlier, 1 the same month, 1 later
2203.17003 (2022-03)
Equivariant Diffusion for Molecule Generation in 3D
2305.01140 (2023-05)
Geometric Latent Diffusion Models for 3D Molecule Generation
2410.06502 (2024-10)
Chemistry-Inspired Diffusion with Non-Differentiable Guidance
and 1 more not named here
compute_n_nodes_dict 2 other papers
1 dated earlier, 1 later
2203.17003 (2022-03)
Equivariant Diffusion for Molecule Generation in 3D
2503.02918 (2025-03)
Straight-Line Diffusion Model for Efficient 3D Molecular Generation
get_dataset_info 2 other papers
2 dated earlier
2304.06174 (2023-04)
Accurate transition state generation with an object-aware equivariant elementary reaction diffusion model
2305.01140 (2023-05)
Geometric Latent Diffusion Models for 3D Molecule Generation
load_model 1 other paper
1 dated earlier
2305.01140 (2023-05)
Geometric Latent Diffusion Models for 3D Molecule Generation
random_rotation 1 other paper
1 dated earlier
2305.01140 (2023-05)
Geometric Latent Diffusion Models for 3D Molecule Generation

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

We propose SYMDIFF, a method for constructing equivariant diffusion models using the framework of stochastic symmetrisation. SYMDIFF resembles a learned data augmentation that is deployed at sampling time, and is lightweight, computationally efficient, and easy to implement on top of arbitrary off-the-shelf models. In contrast to previous work, SYMDIFF typically does not require any neural network components that are intrinsically equivariant, avoiding the need for complex parameterisations or the use of higher-order geometric features. Instead, our method can leverage highly scalable modern architectures as drop-in replacements for these more constrained alternatives. We show that this additional flexibility yields significant empirical benefit for E(3)-equivariant molecular generation. To the best of our knowledge, this is the first application of symmetrisation to generative modelling, suggesting its potential in this domain more generally. Published as a conference paper at ICLR 2025 tractable optimisation objective. Our model is stochastically E(3)-equivariant overall without needing any intrinsically E(3)-equivariant neural networks as subcomponents. We also sketch how to extend SYMDIFF to score and flow-based generative models (Song et al., 2020;Lipman et al., 2022). To validate our framework, we implemented SYMDIFF for de novo molecular generation, and evaluated it as a drop-in replacement for the E(3)-equivariant diffusion of Hoogeboom et al. ( 2022), which relies on intrinsically equivariant neural networks. In contrast, our model is able to leverage highly scalable off-the-shelf architectures such as Diffusion Transformers (Peebles & Xie, 2023) for all of its subcomponents. We demonstrate this leads to significantly improved empirical performance for both the QM9 and GEOM-Drugs datasets. We provide here an overview of the underlying theory behind equivariant diffusion modelling. This theory is most conveniently developed in terms of Markov kernels, whose definition we recall first. Markov kernels At a high level, a Markov kernel k : X → Y may be thought of as a conditional distribution or stochastic map that, when given an input x ∈ X , produces a random output in Y with distribution k(dy|x). For example, given a function f : X × E → Y and a random element η of E, there is a Markov kernel k : X → Y for which k(dy|x) is the distribution of f (x, η) 1 . As a special case, every deterministic function f : X → Y may be thought of as a Markov kernel X → Y also. When k(dy|x) has a density (or likelihood), we will denote this by k(y|x), although we note that we can still reason about Markov kernels even when they do not admit a likelihood in this sense.

For agents

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

get_harvested_code_for_paper("2410.06262")
get_code_for_paper("2410.06262")
have("2410.06262")

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

curl https://syntology.ai/api/ran/2410.06262.json

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

[![Syntology run record](https://syntology.ai/api/ran/2410.06262.svg)](https://syntology.ai/paper/2410.06262)

Connect an agent — have() is free.