DD-Ranking: Rethinking the Evaluation of Dataset Distillation

| Documentation | Leaderboard | Paper | Twitter/X | Developer Slack |

---

Latest News 🔥

[Latest] We have released the v0.2.0 version of DD-Ranking. Please install the latest version via pip install ddranking==0.2.0 or pip install ddranking --upgrade.

Unfold to see more details.

• [2025/06] We have released the v0.2.0 version of DD-Ranking. Please install the latest version via pip install ddranking==0.2.0 or pip install ddranking --upgrade.

• [2025/02] We have fixed some bugs and released a new version of DD-Ranking. Please update your package via pip install ddranking==0.1.4 or pip install ddranking --upgrade.

• [2025/01] Our PyPI package is officially released! Users can now install DD-Ranking via pip install ddranking.

• [2024/12/28] We officially released DD-Ranking! DD-Ranking provides us a new benchmark decoupling the impacts from knowledge distillation and data augmentation.

---

Motivation: DD Lacks an Evaluation Benchmark

Unfold to see more details.

Dataset Distillation (DD) aims to condense a large dataset into a much smaller one, which allows a model to achieve comparable performance after training on it. DD has gained extensive attention since it was proposed. With some foundational methods such as DC, DM, and MTT, various works have further pushed this area to a new standard with their novel designs.

Notebaly, more and more methods are transitting from "hard label" to "soft label" in dataset distillation, especially during evaluation. Hard labels are categorical, having the same format of the real dataset. Soft labels are outputs of a pre-trained teacher model. Recently, Deng et al., pointed out that "a label is worth a thousand images". They showed analytically that soft labels are exetremely useful for accuracy improvement.

However, since the essence of soft labels is knowledge distillation, we find that when applying the same evaluation method to randomly selected data, the test accuracy also improves significantly (see the figure above).

This makes us wonder: Can the test accuracy of the model trained on distilled data reflect the real informativeness of the distilled data?

We summaize the evaluation configurations of existing works in the following table, with different colors highlighting different values for each configuration. As can be easily seen, the evaluation configurations are diverse, leading to unfairness of using only test accuracy to demonstrate one's performance. Among these inconsistencies, two critical factors significantly undermine the fairness of current evaluation protocols: label representation (including the corresponding loss function) and data augmentation techniques.

Motivated by this, we propose DD-Ranking, a new benchmark for DD evaluation. DD-Ranking provides a fair evaluation scheme for DD methods that can decouple the impacts from knowledge distillation and data augmentation to reflect the real informativeness of the distilled data.

Introduction

Unfold to see more details.

DD-Ranking (DD, *i.e.*, Dataset Distillation) is an integrated and easy-to-use benchmark for dataset distillation. It aims to provide a fair evaluation scheme for DD methods that can decouple the impacts from knowledge distillation and data augmentation to reflect the real informativeness of the distilled data.

Benchmark

Revisit the original goal of dataset distillation:

The idea is to synthesize a small number of data points that do not need to come from the correct data distribution, but will, when given to the learning algorithm as training data, approximate the model trained on the original data. (Wang et al., 2020)

Label-Robust Score (LRS)

For the label representation, we introduce the Label-Robust Score (LRS) to evaluate the informativeness of the synthesized data using the following two aspects:

1. The degree to which the real dataset is recovered under hard labels (hard label recovery): $\text{HLR}=\text{Acc.}{\text{real-hard}}-\text{Acc.}{\text{syn-hard}}$.

2. The improvement over random selection when using personalized evaluation methods (improvement over random): $\text{IOR}=\text{Acc.}{\text{syn-any}}-\text{Acc.}{\text{rdm-any}}$. $\text{Acc.}$ is the accuracy of models trained on different samples. Samples' marks are as follows:

• $\text{real-hard}$: Real dataset with hard labels;
• $\text{syn-hard}$: Synthetic dataset with hard labels;
• $\text{syn-any}$: Synthetic dataset with personalized evaluation methods (hard or soft labels);
• $\text{rdm-any}$: Randomly selected dataset (under the same compression ratio) with the same personalized evaluation methods.

LRS is defined as a weight sum of $\text{IOR}$ and $-\text{HLR}$ to rank different methods: $\alpha = w\text{IOR}-(1-w)\text{HLR}, \quad w \in [0, 1]$. Then, the LRS is normalized to $[0, 1]$ as follows: $\text{LRS} = (e^{\alpha}-e^{-1}) / (e - e^{-1})$

By default, we set $w = 0.5$ on the leaderboard, meaning that both $\text{IOR}$ and $\text{HLR}$ are equally important. Users can adjust the weights to emphasize one aspect on the leaderboard.

Augmentation-Robust Score (ARS)

To disentangle data augmentation’s impact, we introduce the augmentation-robust score (ARS) which continues to leverage the relative improvement over randomly selected data. Specifically, we first evaluate synthetic data and a randomly selected subset under the same setting to obtain $\text{Acc.}{\text{syn-aug}}$ and $\text{Acc.}{\text{rdm-aug}}$ (same as IOR). Next, we evaluate both synthetic data and random data again without the data augmentation, and results are denoted as $\text{Acc.}{\text{syn-naug}}$ and $\text{Acc.}{\text{rdm-naug}}$. Both differences, $\text{accsyn-aug} - \text{accrdm-aug}$ and $\text{accsyn-naug} - \text{accrdm-naug}$, are positively correlated to the real informativeness of the distilled dataset.

ARS is a weighted sum of the two differences: $\beta = \gamma(\text{accsyn-aug} - \text{accrdm-aug}) + (1 - \gamma)(\text{accsyn-naug} - \text{accrdm-naug})$, and normalized to $[0, 1]$ similarly.

Overview

DD-Ranking is integrated with:

• Multiple strategies of using soft labels in existing works;
• Commonly used data augmentation methods in existing works;
• Commonly used model architectures in existing works.

DD-Ranking has the following features:

• Fair Evaluation: DD-Ranking provides a fair evaluation scheme for DD methods that can decouple the impacts from knowledge distillation and data augmentation to reflect the real informativeness of the distilled data.
• Easy-to-use: DD-Ranking provides a unified interface for dataset distillation evaluation.
• Extensible: DD-Ranking supports various datasets and models.
• Customizable: DD-Ranking supports various data augmentations and soft label strategies.

DD-Ranking currently includes the following datasets and methods (categorized by hard/soft label). Our replication of the following baselines can be found at the methods branch. Evaluation results can be found in the leaderboard and evaluation configurations can be found at the eval branch.

Supported Dataset	Evaluated Hard Label Methods	Evaluated Soft Label Methods
CIFAR-10	DC	DATM
CIFAR-100	DSA	SRe2L
TinyImageNet	DM	RDED
ImageNet1K	MTT	D4M
	DataDAM	EDF
		CDA
		DWA
		EDC
		G-VBSM

Tutorial

Install DD-Ranking with pip or from source:

Installation

From pip

pip install ddranking

From source

python setup.py install

Quickstart

Below is a step-by-step guide on how to use our ddranking. This demo is based on LRS on soft labels (source code can be found in demo_lrs_soft.py). You can find LRS on hard labels in demo_lrs_hard.py and ARS in demo_aug.py. DD-Ranking supports multi-GPU Distributed evaluation. You can simply use torchrun to launch the evaluation.

Step1: Intialize a soft-label metric evaluator object. Config files are recommended for users to specify hyper-parameters. Sample config files are provided here.

from ddranking.metrics import LabelRobustScoreSoft
from ddranking.config import Config

config = Config.from_file("./configs/Demo_LRS_Soft_Label.yaml")
lrs_soft_metric = LabelRobustScoreSoft(config)

You can also pass keyword arguments.

device = "cuda"
method_name = "DATM"                    # Specify your method name
ipc = 10                                # Specify your IPC
dataset = "CIFAR100"                     # Specify your dataset name
syn_data_dir = "./data/CIFAR100/IPC10/"  # Specify your synthetic data path
real_data_dir = "./datasets"            # Specify your dataset path
model_name = "ConvNet-3"                # Specify your model name
teacher_dir = "./teacher_models"		# Specify your path to teacher model chcekpoints
teacher_model_names = ["ConvNet-3"]      # Specify your teacher model names
im_size = (32, 32)                      # Specify your image size
dsa_params = {                          # Specify your data augmentation parameters
    "prob_flip": 0.5,
    "ratio_rotate": 15.0,
    "saturation": 2.0,
    "brightness": 1.0,
    "contrast": 0.5,
    "ratio_scale": 1.2,
    "ratio_crop_pad": 0.125,
    "ratio_cutout": 0.5
}
random_data_format = "tensor"              # Specify your random data format (tensor or image)
random_data_path = "./random_data"          # Specify your random data path
save_path = f"./results/{dataset}/{model_name}/IPC{ipc}/dm_hard_scores.csv"

""" We only list arguments that usually need specifying"""
lrs_soft_metric = LabelRobustScoreSoft(
    dataset=dataset,
    real_data_path=real_data_dir, 
    ipc=ipc,
    model_name=model_name,
    soft_label_criterion='sce',  # Use Soft Cross Entropy Loss
    soft_label_mode='S',         # Use one-to-one image to soft label mapping
    loss_fn_kwargs={'temperature': 1.0, 'scale_loss': False},
    data_aug_func='dsa',         # Use DSA data augmentation
    aug_params=dsa_params,       # Specify dsa parameters
    im_size=im_size,
    random_data_format=random_data_format,
    random_data_path=random_data_path,
    stu_use_torchvision=False,
    tea_use_torchvision=False,
    teacher_dir=teacher_dir,
    teacher_model_names=teacher_model_names,
    num_eval=5,
    device=device,
    dist=True,
    save_path=save_path
)

For detailed explanation for hyper-parameters, please refer to our documentation.

Step 2: Load your synthetic data, labels (if any), and learning rate (if any).

syn_images = torch.load('/your/path/to/syn/images.pt')
# You must specify your soft labels if your soft label mode is 'S'
soft_labels = torch.load('/your/path/to/syn/labels.pt')
syn_lr = torch.load('/your/path/to/syn/lr.pt')

Step 3: Compute the metric.

lrs_soft_metric.compute_metrics(image_tensor=syn_images, soft_labels=soft_labels, syn_lr=syn_lr)
# alternatively, you can specify the image folder path to compute the metric
lrs_soft_metric.compute_metrics(image_path='./your/path/to/syn/images', soft_labels=soft_labels, syn_lr=syn_lr)

The following results will be printed and saved to save_path:

• HLR mean: The mean of hard label recovery over num_eval runs.
• HLR std: The standard deviation of hard label recovery over num_eval runs.
• IOR mean: The mean of improvement over random over num_eval runs.
• IOR std: The standard deviation of improvement over random over num_eval runs.
• LRS mean: The mean of Label-Robust Score over num_eval runs.
• LRS std: The standard deviation of Label-Robust Score over num_eval runs.

Check out our documentation to learn more.

Contributing

Feel free to submit grades to update the DD-Ranking list. We welcome and value any contributions and collaborations. Please check out CONTRIBUTING.md for how to get involved.

License

DD-Ranking is released under the MIT License. See LICENSE for more details.

Reference

If you find DD-Ranking useful in your research, please consider citing the following paper:

@misc{li2025ddrankingrethinkingevaluationdataset,
      title={DD-Ranking: Rethinking the Evaluation of Dataset Distillation}, 
      author={Zekai Li and Xinhao Zhong and Samir Khaki and Zhiyuan Liang and Yuhao Zhou and Mingjia Shi and Ziqiao Wang and Xuanlei Zhao and Wangbo Zhao and Ziheng Qin and Mengxuan Wu and Pengfei Zhou and Haonan Wang and David Junhao Zhang and Jia-Wei Liu and Shaobo Wang and Dai Liu and Linfeng Zhang and Guang Li and Kun Wang and Zheng Zhu and Zhiheng Ma and Joey Tianyi Zhou and Jiancheng Lv and Yaochu Jin and Peihao Wang and Kaipeng Zhang and Lingjuan Lyu and Yiran Huang and Zeynep Akata and Zhiwei Deng and Xindi Wu and George Cazenavette and Yuzhang Shang and Justin Cui and Jindong Gu and Qian Zheng and Hao Ye and Shuo Wang and Xiaobo Wang and Yan Yan and Angela Yao and Mike Zheng Shou and Tianlong Chen and Hakan Bilen and Baharan Mirzasoleiman and Manolis Kellis and Konstantinos N. Plataniotis and Zhangyang Wang and Bo Zhao and Yang You and Kai Wang},
      year={2025},
      eprint={2505.13300},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2505.13300}, 
}