SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2302.04867 · NeurIPS · 2023

UniPC: A Unified Predictor-Corrector Framework for Fast Sampling of Diffusion Models

Yongming Rao, Wenliang Zhao, Jiwen Lu, Jie Zhou, Lujia Bai

arXiv · PDF · Open in the Atlas

Code that ran

6 of 8 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (1 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 5 failures of our instrument; 2 no recorded run); 2 of them were also run on the host, not in the jail, by the peer-agreement lane that set their level, 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/2302.04867.json · badge: /api/ran/2302.04867.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
wl-zhao/unipc canonical (classifier label)
classifier basis: classifier_2025
5 ran · 0 no recorded run
wl-zhao/UniPC canonical (classifier label)
classifier basis: classifier_2025
1 ran · 2 no recorded run
FunctionStatusWhere it lives
interpolate_fn Ran
ran; our contract draft was wrong, not the code
wl-zhao/UniPC/example/score_sde_pytorch/uni_pc.py
code served (permissive licence) · get_code("39c68b6295eaca2f")
How it ran

Ran in Syntology's jail 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, 1) as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-04T23:19:46Z). 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)

# x: PyTorch tensor with shape [N, C], where N is batch size, C is number of channels
x = torch.randn(2, 1)

# xp: PyTorch tensor with shape [C, K], where K is number of keypoints
xp = torch.linspace(0, 1, 4).unsqueeze(0).expand(1, -1)

# yp: PyTorch tensor with shape [C, K]
yp = torch.linspace(0, 2, 4).unsqueeze(0).expand(1, -1)

ARGS = {
    "x": x,
    "xp": xp,
    "yp": yp,
}
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, 1) — shape (2, 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
interpolate_fn Ran
ran; our contract draft was wrong, not the code
wl-zhao/unipc/example/score_sde_pytorch/uni_pc.py
code served (permissive licence) · get_code("6a35b62fbc80f70a")
How it ran

Ran in Syntology's jail 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, 1) as a later run recorded it. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-05T00:15:28Z). 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)

# x: PyTorch tensor with shape [N, C], where N is batch size, C is number of channels
x = torch.randn(2, 1)

# xp: PyTorch tensor with shape [C, K], where K is number of keypoints
xp = torch.linspace(0, 1, 4).unsqueeze(0).repeat(1, 1)  # shape [1, 4]

# yp: PyTorch tensor with shape [C, K]
yp = torch.linspace(0, 2, 4).unsqueeze(0).repeat(1, 1)  # shape [1, 4]

ARGS = {
    "x": x,
    "xp": xp,
    "yp": yp,
}
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, 1) — shape (2, 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
UniPC Ran
ran on a synthesized input · constructed an object; no result was computed
wl-zhao/unipc/uni_pc.py
code served (permissive licence) · get_code("bbcc7bee81a99149")
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, UniPC: 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)

# Create a minimal noise schedule class
class NoiseSchedule:
    def __init__(self):
        pass
    
    def marginal_alpha(self, t):
        # Return alpha values based on time t
        # t can be a scalar or tensor
        if isinstance(t, torch.Tensor):
            return torch.sqrt(1.0 - t ** 2)
        return (1.0 - t ** 2) ** 0.5
    
    def marginal_std(self, t):
        # Return sigma values based on time t
        if isinstance(t, torch.Tensor):
            return t
        return t
    
    def marginal_lambda(self, t):
        # Return lambda (log SNR) values
        alpha = self.marginal_alpha(t)
        sigma = self.marginal_std(t)
        # lambda = log(alpha^2 / sigma^2)
        return torch.log(alpha ** 2 / (sigma ** 2 + 1e-8))
    
    def inverse_lambda(self, lambda_t):
        # Inverse of marginal_lambda
        # Given lambda, return t
        # This is a simplified inverse
        return torch.exp(-lambda_t / 2)

# Create a minimal model function
class DiffusionModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t):
        # x: [batch, 4]
        # t: [batch]
        # Return noise prediction or data prediction
        return self.linear(x)

# Instantiate the model
model = DiffusionModel()

# Create a wrapper function that matches the expected signature
def model_fn(x, t):
    return model(x, t)

# Instantiate noise schedule
noise_schedule = NoiseSchedule()

# Create the ARGS dictionary
ARGS = {
    'model_fn': model_fn,
    'noise_schedule': noise_schedule,
    'algorithm_type': 'data_prediction',
    'correcting_x0_fn': None,
    'correcting_xt_fn': None,
    'thresholding_max_val': 1.0,
    'dynamic_thresholding_ratio': 0.995,
    'variant': 'bh1'
}
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 UniPC — 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 UniPC 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
expand_dims Ran
ran; our contract draft was wrong, not the code
wl-zhao/unipc/example/score_sde_pytorch/uni_pc.py
code served (permissive licence) · get_code("e6110366588c5c65")
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, 1, 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

torch.manual_seed(0)

# Create a small 1D tensor with shape [N]
v = torch.randn(2)

# Create a dims parameter - total number of dimensions desired
dims = 4

ARGS = {
    "v": v,
    "dims": dims,
}
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, 1, 1, 1) — shape (2, 1, 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
model_wrapper Ran
ran; our contract draft was wrong, not the code
wl-zhao/unipc/example/score_sde_pytorch/uni_pc.py
code served (permissive licence) · get_code("64098eb6704a58dd")
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
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal noise schedule
class NoiseSchedule:
    def __init__(self):
        self.schedule = 'discrete'
        self.total_N = 1000
    
    def marginal_alpha(self, t):
        return torch.sqrt(1 - t)
    
    def marginal_std(self, t):
        return torch.sqrt(t)

# Create a minimal model that accepts (x, t) or (x, t, cond)
class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t, cond=None):
        # x: (batch, 4), t: scalar or tensor, cond: optional
        return self.linear(x)

# Create a minimal classifier for classifier guidance
class SimpleClassifier(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 1)
    
    def forward(self, x, t, condition, **kwargs):
        # Returns log probability
        return self.linear(x).sum(dim=-1, keepdim=True)

# Build the arguments
model = SimpleModel()
noise_schedule = NoiseSchedule()
model_type = 'noise'
model_kwargs = {}
guidance_type = 'uncond'
condition = None
unconditional_condition = None
guidance_scale = 1.0
classifier_fn = None
classifier_kwargs = {}

ARGS = {
    'model': model,
    'noise_schedule': noise_schedule,
    'model_type': model_type,
    'model_kwargs': model_kwargs,
    'guidance_type': guidance_type,
    'condition': condition,
    'unconditional_condition': unconditional_condition,
    'guidance_scale': guidance_scale,
    'classifier_fn': classifier_fn,
    'classifier_kwargs': classifier_kwargs,
}
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
model_wrapper Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
wl-zhao/unipc/example/stable-diffusion/ldm/models/diffusion/uni_pc/uni_pc.py
code served (permissive licence) · get_code("3552d38797b670d4")
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
import torch.nn as nn

torch.manual_seed(0)

# Create a minimal noise schedule object
class NoiseScheduleVP:
    def __init__(self):
        self.T = 1.0
    
    def marginal_log_mean_coeff(self, t):
        return -0.5 * t
    
    def marginal_std(self, t):
        return torch.sqrt(1.0 - torch.exp(2.0 * self.marginal_log_mean_coeff(t)))
    
    def inverse_lambda(self, lambda_t):
        return -2.0 * lambda_t

# Create a minimal diffusion model
class MinimalDiffusionModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 4)
    
    def forward(self, x, t, cond=None, **kwargs):
        # x shape: [batch, 4]
        # t shape: [batch] or scalar
        # Returns noise prediction of same shape as x
        return self.linear(x)

# Create a minimal classifier function
def minimal_classifier_fn(x, t, cond, **kwargs):
    # Returns logits
    batch_size = x.shape[0]
    return torch.randn(batch_size, 10)

# Create sample tensors
batch_size = 2
x_dim = 4

sample_x = torch.randn(batch_size, x_dim)
condition = torch.randn(batch_size, 3)
unconditional_condition = torch.zeros(batch_size, 3)

# Instantiate model and noise schedule
model = MinimalDiffusionModel()
noise_schedule = NoiseScheduleVP()

ARGS = {
    'model': model,
    'noise_schedule': noise_schedule,
    'model_type': 'noise',
    'model_kwargs': {'extra_param': 1},
    'guidance_type': 'uncond',
    'condition': condition,
    'unconditional_condition': unconditional_condition,
    'guidance_scale': 1.0,
    'classifier_fn': minimal_classifier_fn,
    'classifier_kwargs': {'classifier_param': 0.5}
}
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
expand_dims Not yet run
unverified (harvested, no recorded run)
wl-zhao/UniPC/example/score_sde_pytorch/uni_pc.py
code served (permissive licence) · get_code("292d0a3a80570d32")
How it ran

no invocation recorded. this sample carries JSON test cases instead of a setup block; get_reference_implementation serves them with the level they back.

model_wrapper Not yet run
unverified (harvested, no recorded run)
wl-zhao/UniPC/example/score_sde_pytorch/uni_pc.py
code served (permissive licence) · get_code("812cebe843fafdb2")
How it ran

no invocation recorded. this sample carries JSON test cases instead of a setup block; get_reference_implementation serves them with the level they back.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 4 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 4 later) and 1 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
expand_dims
brick of this paper's assembly: helper
39 other papers
6 dated earlier, 33 later · 4 only as an assembly brick
2112.10752 (2021-12)
High-Resolution Image Synthesis with Latent Diffusion Models
2202.09778 (2022-02)
Pseudo Numerical Methods for Diffusion Models on Manifolds
2206.00927 (2022-06)
DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps
and 36 more not named here
interpolate_fn 34 other papers
4 dated earlier, 30 later
2112.10752 (2021-12)
High-Resolution Image Synthesis with Latent Diffusion Models
2211.01095 (2022-11)
DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Models
2212.09478 (2022-12)
MM-Diffusion: Learning Multi-Modal Diffusion Models for Joint Audio and Video Generation
and 31 more not named here
model_wrapper 4 other papers
1 dated earlier, 3 later
2211.01095 (2022-11)
DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Models
2306.02018 (2023-06)
VideoComposer: Compositional Video Synthesis with Motion Controllability
2310.13268 (2023-10)
DPM-Solver-v3: Improved Diffusion ODE Solver with Empirical Model Statistics
and 1 more not named here
model_wrapper 1 other paper
1 later
2409.03755 (2024-09)
DC-Solver: Improving Predictor-Corrector Diffusion Sampler via Dynamic Compensation

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

Diffusion probabilistic models (DPMs) have demonstrated a very promising ability in high-resolution image synthesis. However, sampling from a pre-trained DPM is time-consuming due to the multiple evaluations of the denoising network, making it more and more important to accelerate the sampling of DPMs. Despite recent progress in designing fast samplers, existing methods still cannot generate satisfying images in many applications where fewer steps (e.g., <10) are favored. In this paper, we develop a unified corrector (UniC) that can be applied after any existing DPM sampler to increase the order of accuracy without extra model evaluations, and derive a unified predictor (UniP) that supports arbitrary order as a byproduct. Combining UniP and UniC, we propose a unified predictor-corrector framework called UniPC for the fast sampling of DPMs, which has a unified analytical form for any order and can significantly improve the sampling quality over previous methods, especially in extremely few steps. We evaluate our methods through extensive experiments including both unconditional and conditional sampling using pixel-space and latent-space DPMs. Our UniPC can achieve 3.87 FID on CIFAR10 (unconditional) and 7.51 FID on ImageNet 256×256 (conditional) with only 10 function evaluations. Code is available at https://github.com/wl-zhao/UniPC.

For agents

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

get_harvested_code_for_paper("2302.04867")
get_code_for_paper("2302.04867")
have("2302.04867")

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

curl https://syntology.ai/api/ran/2302.04867.json

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

[![Syntology run record](https://syntology.ai/api/ran/2302.04867.svg)](https://syntology.ai/paper/2302.04867)

Connect an agent — have() is free.