SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2011.13456 · ICLR · 2021

Score-Based Generative Modeling through Stochastic Differential Equations

Abhishek Kumar, Ben Poole, Yang Song, Stefano Ermon, Diederik Kingma, Jascha Sohl-Dickstein

arXiv · PDF · Open in the Atlas

Code that ran

15 of 23 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (4 of them constructed an object rather than computing a result; 2 honoured the contract we drafted, 0 violated it and 6 have no contract verdict; 7 failures of our instrument; 8 no recorded run), measured 2026-11-21; 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/2011.13456.json · badge: /api/ran/2011.13456.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
yang-song/score_sde canonical (classifier label)
classifier basis: classifier_2025
3 ran · 4 no recorded run
intuitive-robots/beso community
Papers with Code archive
3 ran · 2 no recorded run
CW-Huang/sdeflow-light community
Papers with Code archive
3 ran · 0 no recorded run
copy not recorded community
not in the paper's declared repositories
2 ran · 0 no recorded run
liaopeiyuan/pndm community
Papers with Code archive
1 ran · 2 no recorded run
g4vrel/sde_ddpm community
Papers with Code archive
1 ran · 0 no recorded run
luping-liu/PNDM community
Papers with Code archive
1 ran · 0 no recorded run
yanfeng-yang-0316/cdcit community
Papers with Code archive
1 ran · 0 no recorded run
FunctionStatusWhere it lives
EulerMaruayamaPredictor Ran
ran on a synthesized input
g4vrel/sde_ddpm/sde.py
pointer only (licence: NONE) · get_code("a255c2fd61329566")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, g4vrel/sde_ddpm 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.

NoneCorrector Ran
ran on a synthesized input · constructed an object; no result was computed
yang-song/score_sde/sampling.py
code served (permissive licence) · get_code("6f447a2b95b54845")
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 an instance of its own class, NoneCorrector: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

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

torch.manual_seed(0)

# Define a minimal SDE class
class MinimalSDE:
    def __init__(self):
        pass

# Define a minimal score function (nn.Module)
class MinimalScoreFn(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        return self.linear(x)

# Create instances
sde = MinimalSDE()
score_fn = MinimalScoreFn()
snr = 0.1
n_steps = 10

ARGS = {
    'sde': sde,
    'score_fn': score_fn,
    'snr': snr,
    'n_steps': n_steps
}
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 NoneCorrector — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a NoneCorrector 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
NonePredictor Ran
ran on a synthesized input · constructed an object; no result was computed
yang-song/score_sde/sampling.py
code served (permissive licence) · get_code("6fe4dbf5838e4df8")
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 an instance of its own class, NonePredictor: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

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

torch.manual_seed(0)

# Create a minimal SDE class with reverse method
class MockSDE:
    def __init__(self):
        pass
    
    def reverse(self, score_fn, probability_flow):
        return MockReverseSDE()

class MockReverseSDE:
    pass

# Create a minimal score function (nn.Module)
class MockScoreFunction(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        return self.linear(x)

# Set up arguments
sde = MockSDE()
score_fn = MockScoreFunction()
probability_flow = False

ARGS = {
    'sde': sde,
    'score_fn': score_fn,
    'probability_flow': probability_flow
}
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 NonePredictor — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a NonePredictor 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
PIDStepSizeController Ran
ran on a synthesized input
intuitive-robots/beso/beso/agents/diffusion_agents/k_diffusion/gc_sampling.py
code served (permissive licence) · get_code("23c37bd71f8e0dbd")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned bool, ndim 0, 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 math

class PIDStepSizeController:
    """A PID controller for ODE adaptive step size control."""

    def __init__(self, h, pcoeff, icoeff, dcoeff, order=1, accept_safety=0.81, eps=1e-08):
        self.h = h
        self.b1 = (pcoeff + icoeff + dcoeff) / order
        self.b2 = -(pcoeff + 2 * dcoeff) / order
        self.b3 = dcoeff / order
        self.accept_safety = accept_safety
        self.eps = eps
        self.errs = []

    def limiter(self, action):
        return 1 + math.atan(action - 1)

    def propose_step(self, error):
        inv_error = 1 / (float(error) + self.eps)
        if not self.errs:
            self.errs = [inv_error, inv_error, inv_error]
        self.errs[0] = inv_error
        factor = self.errs[0] ** self.b1 * self.errs[1] ** self.b2 * self.errs[2] ** self.b3
        factor = self.limiter(factor)
        accept = factor >= self.accept_safety
        if accept:
            self.errs[2] = self.errs[1]
            self.errs[1] = self.errs[0]
        self.h *= factor
        return accept


INIT_ARGS = {
    'h': 0.1,
    'pcoeff': 0.3,
    'icoeff': 0.2,
    'dcoeff': 0.1,
    'order': 1,
    'accept_safety': 0.81,
    'eps': 1e-08
}

ARGS = {
    'error': 0.05
}

METHOD = 'propose_step'

if __name__ == '__main__':
    controller = PIDStepSizeController(**INIT_ARGS)
    result = controller.propose_step(**ARGS)
    print(f"Accept: {result}")
    print(f"New step size h: {controller.h}")
    print(f"Error history: {controller.errs}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
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 bool, ndim 0, dtype family float — no output shape is stated: observe_signature.py observed this class with its constructor sizes canonicalised where it could, which the node does not record, so its shape is not stated as this setup block's
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: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
VariancePreservingSDE Ran
ran on a synthesized input · constructed an object; no result was computed
CW-Huang/sdeflow-light/lib/sdes.py
code served (permissive licence) · get_code("961b5bdb253bde68")
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 an instance of its own class, VariancePreservingSDE: the run constructed the object and did not call it, so no result was computed. 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)

ARGS = {
    "beta_min": 0.1,
    "beta_max": 20.0,
    "T": 1.0,
    "t_epsilon": 0.001,
}
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 VariancePreservingSDE — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a VariancePreservingSDE 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
VariancePreservingTruncatedSampling Ran
ran on a synthesized input · constructed an object; no result was computed
CW-Huang/sdeflow-light/lib/sdes.py
code served (permissive licence) · get_code("559d174d6d82afd7")
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 an instance of its own class, VariancePreservingTruncatedSampling: the run constructed the object and did not call it, so no result was computed. 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

torch.manual_seed(0)

ARGS = {
    "beta_min": 0.1,
    "beta_max": 20.0,
    "t_epsilon": 0.001,
}
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 VariancePreservingTruncatedSampling — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a VariancePreservingTruncatedSampling 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
default_noise_sampler Ran
ran; our contract draft was wrong, not the code
intuitive-robots/beso/beso/agents/diffusion_agents/k_diffusion/gc_sampling.py
code served (permissive licence) · get_code("0f41c312acba15d8")
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 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 torch

torch.manual_seed(0)

# The function takes parameter 'x' which is used with torch.randn_like(x)
# So x needs to be a tensor
x = torch.randn(2, 4)

ARGS = {
    "x": x
}
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 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
get_ancestral_step Ran
ran, honoured the contract we drafted
intuitive-robots/beso/beso/agents/diffusion_agents/k_diffusion/gc_sampling.py
code served (permissive licence) · get_code("e79707d0454c79c9")
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 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 numpy as np

torch.manual_seed(0)

# Analyze the function:
# - sigma_from: used in arithmetic operations (**, /, -), should be a float or scalar
# - sigma_to: used in arithmetic operations (**, /, -), should be a float or scalar
# - eta: used in conditional (if not eta) and arithmetic operations, should be a float or scalar

# The function performs mathematical operations on noise levels (sigmas)
# These are typically scalar values in diffusion models

ARGS = {
    "sigma_from": 2.0,
    "sigma_to": 1.0,
    "eta": 1.0,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — 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
get_pc_sampler Ran
ran; our contract draft was wrong, not the code
yang-song/score_sde/sampling.py
code served (permissive licence) · get_code("43fddf16d4b421ac")
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 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 jax
import jax.numpy as jnp
import jax.random as random
import abc
import functools

# Mock mutils module since it's referenced in the code
class mutils:
    @staticmethod
    def get_score_fn(sde, model, params_ema, model_state, train=False, continuous=False):
        """Mock score function getter"""
        def score_fn(x, t):
            return jnp.zeros_like(x)
        return score_fn

# Mock SDE class
class MockSDE:
    def __init__(self):
        self.T = 1.0
    
    def reverse(self, score_fn, probability_flow=False):
        """Return a reversed SDE"""
        return self

# Mock model class
class MockModel:
    def __init__(self):
        self.params_ema = {}
        self.model_state = {}

# Mock state class
class MockState:
    def __init__(self):
        self.params_ema = {}
        self.model_state = {}

# Mock inverse scaler
def mock_inverse_scaler(x):
    return x

# Create minimal instances
sde = MockSDE()
model = MockModel()
shape = (2, 4)  # batch_size=2, feature_dim=4
predictor = None  # Will use NonePredictor
corrector = None  # Will use NoneCorrector
inverse_scaler = mock_inverse_scaler
snr = 0.1
n_steps = 1
probability_flow = False
continuous = False
denoise = True
eps = 0.001

ARGS = {
    'sde': sde,
    'model': model,
    'shape': shape,
    'predictor': predictor,
    'corrector': corrector,
    'inverse_scaler': inverse_scaler,
    'snr': snr,
    'n_steps': n_steps,
    'probability_flow': probability_flow,
    'continuous': continuous,
    'denoise': denoise,
    'eps': eps
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
jax — 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
register_corrector Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from yang-song/score_sde_pytorch
pointer only · get_code("a05b1d0cf3f508b4")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

register_predictor Ran
ran; our contract draft was wrong, not the code
this paper's copy was not recorded; identical code first harvested from yang-song/score_sde_pytorch
pointer only · get_code("473066ddf62a075a")
How it ran

The body of this sample is withheld: this paper's copy of the code was not recorded (its link names no repository or source path), so no licence was read for it. 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.

runge_kutta Ran
ran; our contract draft was wrong, not the code
luping-liu/PNDM/runner/method.py
code served (permissive licence) · get_code("8e6cd2ebb4cb3161")
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 4, dtype family float, shape (2, 4, 1, 1) 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)

# Define a simple model that takes x and t and returns a tensor of the same shape as x
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        # x shape: (batch, channels, height, width) = (2, 4, 1, 1)
        # t shape: (batch,)
        # Return noise prediction of same shape as x
        batch_size = x.shape[0]
        # Flatten x, process, reshape back
        x_flat = x.view(batch_size, -1)
        out = self.linear(x_flat)
        return out.view_as(x)

# Create model instance
model = SimpleModel()

# Create input tensor x: (batch=2, channels=4, height=1, width=1)
x = torch.randn(2, 4, 1, 1)

# Create time list with 3 timesteps
t_list = [
    torch.tensor([100.0, 100.0]),  # t_list[0]
    torch.tensor([50.0, 50.0]),    # t_list[1]
    torch.tensor([10.0, 10.0])     # t_list[2]
]

# Create alphas_cump: cumulative product of alphas
# Need to index up to max(t_list) + 1, so create enough values
# alphas_cump should be indexed by t.long() + 1, so we need at least 102 values
alphas_cump = torch.linspace(0.9, 0.1, 102)

# Initialize ets as empty list
ets = []

ARGS = {
    "x": x,
    "t_list": t_list,
    "model": model,
    "alphas_cump": alphas_cump,
    "ets": ets
}
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 4, dtype family float, shape (2, 4, 1, 1) — shape (2, 4, 1, 1) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
sample_from_diff Ran
ran; our contract draft was wrong, not the code
yanfeng-yang-0316/cdcit/model.py
pointer only (licence: NONE) · get_code("c915fab99cb196da")
How it ran

The body of this sample is withheld: GitHub detected no licence in this paper's repository, yanfeng-yang-0316/cdcit 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.

sample_vp_truncated_q Ran
ran; our contract draft was wrong, not the code
CW-Huang/sdeflow-light/lib/sdes.py
code served (permissive licence) · get_code("4675c27602b677ef")
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 = {
    "shape": (2, 4),
    "beta_min": 0.1,
    "beta_max": 20.0,
    "t_epsilon": 0.001,
    "T": 1.0,
}
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, 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
transfer Ran
ran, honoured the contract we drafted
liaopeiyuan/pndm/runner/method.py
code served (permissive licence) · get_code("e3d0f1667dc4e323")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned Tensor, ndim 4, dtype family float, shape (2, 3, 4, 5) as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# x: input tensor, shape (batch, channels, height, width)
x = torch.randn(2, 3, 4, 5)

# t: current timestep, shape (batch,)
t = torch.tensor([10.0, 15.0])

# t_next: next timestep, shape (batch,)
t_next = torch.tensor([11.0, 16.0])

# et: predicted noise/error, same shape as x
et = torch.randn(2, 3, 4, 5)

# alphas_cump: cumulative product of alphas, must have at least max(t_next.long())+2 elements
# Create a tensor with enough elements for indexing at t_next.long() + 1
alphas_cump = torch.tensor([0.9, 0.85, 0.80, 0.75, 0.70, 0.65, 0.60, 0.55, 0.50, 0.45, 0.40, 0.35, 0.30, 0.25, 0.20, 0.15, 0.10, 0.05])

ARGS = {
    "x": x,
    "t": t,
    "t_next": t_next,
    "et": et,
    "alphas_cump": alphas_cump
}
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 4, dtype family float, shape (2, 3, 4, 5) — shape (2, 3, 4, 5) as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Corrector Not yet run
unverified (harvested, no recorded run)
yang-song/score_sde/sampling.py
code served (permissive licence) · get_code("b5b6009cef6e47d8")
How it ran

no invocation recorded.

DPMSolver Not yet run
unverified (harvested, no recorded run)
intuitive-robots/beso/beso/agents/diffusion_agents/k_diffusion/gc_sampling.py
code served (permissive licence) · get_code("dacabf6cda8ac542")
How it ran

no invocation recorded.

Predictor Not yet run
unverified (harvested, no recorded run)
yang-song/score_sde/sampling.py
code served (permissive licence) · get_code("fdabca225111e3dc")
How it ran

no invocation recorded.

gen_order_2 Not yet run
unverified (harvested, no recorded run)
liaopeiyuan/pndm/runner/method.py
code served (permissive licence) · get_code("61d75983e395ab02")
How it ran

no invocation recorded.

improved_eular Not yet run
unverified (harvested, no recorded run)
liaopeiyuan/pndm/runner/method.py
code served (permissive licence) · get_code("fe36f2c4259802ef")
How it ran

no invocation recorded.

sample_dpm_adaptive Not yet run
unverified (harvested, no recorded run)
intuitive-robots/beso/beso/agents/diffusion_agents/k_diffusion/gc_sampling.py
code served (permissive licence) · get_code("3f6adafb3e149eb3")
How it ran

no invocation recorded.

shared_corrector_update_fn Not yet run
unverified (harvested, no recorded run)
yang-song/score_sde/sampling.py
code served (permissive licence) · get_code("0ef020b95f4bc102")
How it ran

no invocation recorded.

shared_predictor_update_fn Not yet run
unverified (harvested, no recorded run)
yang-song/score_sde/sampling.py
code served (permissive licence) · get_code("8a1938617d189156")
How it ran

no invocation recorded.

Also present in other papers' code

Of 23 samples lifted from this paper's linked repositories, 23 carry a function identity and 0 do not. Of the 23 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 2 later) and 20 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
register_corrector
brick of this paper's assembly: helper
15 other papers
1 dated earlier, 14 later · 2 only as an assembly brick
2006.09011 (2020-06)
Improved Techniques for Training Score-Based Generative Models
2105.14080 (2021-05)
Gotta Go Fast When Generating Data with Score-Based Models
2112.07068 (2021-12)
Score-Based Generative Modeling with Critically-Damped Langevin Diffusion
and 12 more not named here
register_predictor
brick of this paper's assembly: helper
15 other papers
1 dated earlier, 14 later · 2 only as an assembly brick
2006.09011 (2020-06)
Improved Techniques for Training Score-Based Generative Models
2105.14080 (2021-05)
Gotta Go Fast When Generating Data with Score-Based Models
2112.07068 (2021-12)
Score-Based Generative Modeling with Critically-Damped Langevin Diffusion
and 12 more not named here
transfer
brick of this paper's assembly: helper
1 other paper
1 dated earlier · 1 only as an assembly brick
2010.02502 (2020-10)
Denoising Diffusion Implicit 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 2026-11-21.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Creating noise from data is easy; creating data from noise is generative modeling. We present a stochastic differential equation (SDE) that smoothly transforms a complex data distribution to a known prior distribution by slowly injecting noise, and a corresponding reverse-time SDE that transforms the prior distribution back into the data distribution by slowly removing the noise. Crucially, the reverse-time SDE depends only on the time-dependent gradient field (a.k.a., score) of the perturbed data distribution. By leveraging advances in score-based generative modeling, we can accurately estimate these scores with neural networks, and use numerical SDE solvers to generate samples. We show that this framework encapsulates previous approaches in score-based generative modeling and diffusion probabilistic modeling, allowing for new sampling procedures and new modeling capabilities. In particular, we introduce a predictor-corrector framework to correct errors in the evolution of the discretized reverse-time SDE. We also derive an equivalent neural ODE that samples from the same distribution as the SDE, but additionally enables exact likelihood computation, and improved sampling efficiency. In addition, we provide a new way to solve inverse problems with score-based models, as demonstrated with experiments on class-conditional generation, image inpainting, and colorization. Combined with multiple architectural improvements, we achieve record-breaking performance for unconditional image generation on CIFAR-10 with an Inception score of 9.89 and FID of 2.20, a competitive likelihood of 2.99 bits/dim, and demonstrate high fidelity generation of 1024 ˆ1024 images for the first time from a score-based generative model.

For agents

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

get_harvested_code_for_paper("2011.13456")
get_code_for_paper("2011.13456")
have("2011.13456")

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

curl https://syntology.ai/api/ran/2011.13456.json

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

[![Syntology run record](https://syntology.ai/api/ran/2011.13456.svg)](https://syntology.ai/paper/2011.13456)

Connect an agent — have() is free.