Minimizing Trajectory Curvature of ODE-based Generative Models

Authors: Sangyun Lee, Beomsu Kim, Jong Chul Ye

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments show that our method achieves a lower curvature than previous models and, therefore, decreased sampling costs while maintaining competitive performance.
Researcher Affiliation Academia 1Soongsil University 2KAIST. Correspondence to: Jong Chul Ye <jong.ye@kaist.ac.kr>.
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes Code is available at https://github.com/sangyun884/fast-ode.
Open Datasets Yes We further conduct an experiment on the image dataset... MNIST, CIFAR-10, and Celeb AHQ (64 64) datasets... FFHQ 64 64, AFHQ 64 64, and Celeb AHQ 256 256 datasets.
Dataset Splits No The paper uses standard datasets like CIFAR-10, MNIST, FFHQ, AFHQ, and Celeb AHQ, which have predefined splits. However, the paper does not explicitly specify the training/validation/test dataset splits (e.g., percentages, sample counts, or explicit citation for splits) used to reproduce the experiments.
Hardware Specification No The paper does not provide specific hardware details (e.g., exact GPU/CPU models, memory amounts) used for running its experiments.
Software Dependencies No The paper mentions using DDPM++ architecture, Adam optimizer, and Scipy's RK45 method, but it does not specify concrete version numbers for these or other software dependencies.
Experiment Setup Yes Table 5 shows the training and architecture configuration we use in our experiments. (This table includes Iterations, Batch size, Learning rate, LR warm-up steps, EMA decay rate, EMA start steps, Dropout probability, Channel multiplier, Channels per resolution, # of Res Blocks, t range).