Skip to content

About Code release for "Dynamical Diffusion: Learning Temporal Dynamics with Diffusion Models" (ICLR 2025)

Notifications You must be signed in to change notification settings

thuml/dynamical-diffusion

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 

Repository files navigation

dynamical-diffusion

About Code release for "Dynamical Diffusion: Learning Temporal Dynamics with Diffusion Models" (ICLR 2025)

Usage

Training

To train a model (e.g., DyDiff for Turbulence), run:

cd core; python train_turbulence_dydiff.py --config_file models/turbulence/dydiff_ema_cosine_ratio05_1st_mask_svd_lr1e-4.yaml

Note: Before running, we need to update the config file with the root for data and the ckpt_path for VAE model.

Sampling

To sampling with the trained model, run:

cd core; python train_turbulence_dydiff.py --config_file models/turbulence/dydiff_ema_cosine_ratio05_1st_mask_svd_lr1e-4.yaml --resume ${model_ckpt} --test

This will generate samples in the logs directory.

Evaluation

After generating samples, evaluate them using the following command:

python core/evaluation/evaluate_turbulence.py --model_output_root logs/turbulence/dydiff_ema_cosine_ratio05_1st_mask_svd_lr1e-4/output_for_evaluation --i3d_model_path ${pretrained_i3d_model}

About

About Code release for "Dynamical Diffusion: Learning Temporal Dynamics with Diffusion Models" (ICLR 2025)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages