SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2407.01812 · 2024

Equivariant Diffusion Policy

arXiv · PDF · Open in the Atlas

Code that ran

4 of 4 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 0 have no contract verdict; 3 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/2407.01812.json · badge: /api/ran/2407.01812.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
pointW/equidiff canonical (classifier label)
classifier basis: census
4 ran · 0 no recorded run
FunctionStatusWhere it lives
get_image_transform Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pointW/equidiff/equi_diffpo/common/cv2_util.py
code served (permissive licence) · get_code("dde317225c508d01")
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 function. 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 math
import cv2
import numpy as np

# Set up arguments for get_image_transform
ARGS = {
    "input_res": (1280, 720),
    "output_res": (640, 480),
    "bgr_to_rgb": False,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function 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
optimal_row_cols Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
pointW/equidiff/equi_diffpo/common/cv2_util.py
code served (permissive licence) · get_code("091393d927ad734b")
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 int; 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 math
import numpy as np

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

ARGS = {
    "n_cameras": 6,
    "in_wh_ratio": 16/9,
    "max_resolution": (1920, 1080)
}
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, ndim 1, dtype family int — 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
protective_squeeze Ran
ran, violated the contract we drafted
pointW/equidiff/equi_diffpo/codecs/imagecodecs_numcodecs.py
code served (permissive licence) · get_code("116acbede4989d1c")
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 4, dtype family float, shape (2, 4, 3, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy

# Create test cases for protective_squeeze
# The function reshapes arrays, treating the last 3 dimensions as image dims (H, W, C)

# Test case 1: 4D array (batch, H, W, C) - should reshape to (-1, H, W, C)
x_4d = numpy.random.randn(2, 4, 3, 5)  # 2 images of shape (4, 3, 5)

# Test case 2: 3D array (H, W, C) - should stay as (H, W, C)
x_3d = numpy.random.randn(4, 3, 5)

# Test case 3: 5D array (n1, n2, H, W, C) - should reshape to (-1, H, W, C) when n1*n2 > 1
x_5d = numpy.random.randn(2, 3, 4, 3, 5)  # 2*3=6 images

ARGS = {
    "x": x_4d
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 4, dtype family float, shape (2, 4, 3, 5) — shape (2, 4, 3, 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
render_env_video Ran
ran; our contract draft was wrong, not the code
pointW/equidiff/equi_diffpo/common/env_util.py
code served (permissive licence) · get_code("5c09b91544d37535")
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 4, dtype family int, shape (2, 96, 96, 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 cv2
import numpy as np


class MockEnv:
    """Mock environment that supports set_state and render."""
    
    def __init__(self):
        self.state = None
        self.img_shape = (96, 96, 3)
    
    def set_state(self, state):
        """Store the state."""
        self.state = state
    
    def render(self):
        """Return a dummy RGB image."""
        return np.random.randint(0, 256, self.img_shape, dtype=np.uint8)


# Create mock environment
env = MockEnv()

# Create states - list of state arrays (2 timesteps, 4-dim state)
states = [
    np.array([0.1, 0.2, 0.3, 0.4], dtype=np.float32),
    np.array([0.5, 0.6, 0.7, 0.8], dtype=np.float32),
]

# Create actions - 2D actions that will be scaled to image coordinates
actions = np.array([
    [256.0, 256.0],
    [384.0, 384.0],
], dtype=np.float32)

ARGS = {
    "env": env,
    "states": states,
    "actions": actions,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
cv2, numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type ndarray, ndim 4, dtype family int, shape (2, 96, 96, 3) — shape (2, 96, 96, 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

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 distinct functions with an identity, 4 are also present in other papers whose code we hold (4 with at least one paper dated earlier, 0 the same month, 0 later) and 0 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
protective_squeeze 4 other papers
4 dated earlier
2303.04137 (2023-03)
Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
2307.01849 (2023-07)
Crossway Diffusion: Improving Diffusion-based Visuomotor Policy via Self-supervised Learning
2403.03206 (2024-03)
Scaling Rectified Flow Transformers for High-Resolution Image Synthesis
and 1 more not named here
get_image_transform 2 other papers
2 dated earlier
2303.04137 (2023-03)
Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
2406.04806 (2024-06)
Streaming Diffusion Policy: Fast Policy Synthesis with Variable Noise Diffusion Models
optimal_row_cols 2 other papers
2 dated earlier
2303.04137 (2023-03)
Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
2406.04806 (2024-06)
Streaming Diffusion Policy: Fast Policy Synthesis with Variable Noise Diffusion Models
render_env_video 2 other papers
2 dated earlier
2303.04137 (2023-03)
Diffusion Policy: Visuomotor Policy Learning via Action Diffusion
2406.04806 (2024-06)
Streaming Diffusion Policy: Fast Policy Synthesis with Variable Noise Diffusion Models

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

Recent work has shown diffusion models are an effective approach to learning the multimodal distributions arising from demonstration data in behavior cloning. However, a drawback of this approach is the need to learn a denoising function, which is significantly more complex than learning an explicit policy. In this work, we propose Equivariant Diffusion Policy, a novel diffusion policy learning method that leverages domain symmetries to obtain better sample efficiency and generalization in the denoising function. We theoretically analyze the $\mathrm{SO}(2)$ symmetry of full 6-DoF control and characterize when a diffusion model is $\mathrm{SO}(2)$-equivariant. We furthermore evaluate the method empirically on a set of 12 simulation tasks in MimicGen, and show that it obtains a success rate that is, on average, 21.9% higher than the baseline Diffusion Policy. We also evaluate the method on a real-world system to show that effective policies can be learned with relatively few training samples, whereas the baseline Diffusion Policy cannot.

For agents

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

get_harvested_code_for_paper("2407.01812")
get_code_for_paper("2407.01812")
have("2407.01812")

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

curl https://syntology.ai/api/ran/2407.01812.json

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

[![Syntology run record](https://syntology.ai/api/ran/2407.01812.svg)](https://syntology.ai/paper/2407.01812)

Connect an agent — have() is free.