|
Download SimDataSystems/readme.md from C3GroupCornell/RC-timescale-data: direct link, hf CLI and curl.
- Browser
- Download file 822 Bytes
-
https://huggingface.co/datasets/C3GroupCornell/RC-timescale-data/resolve/main/SimDataSystems/readme.md
- Command line
-
hf download hf://datasets/C3GroupCornell/RC-timescale-data/SimDataSystems/readme.md
-
curl -L -o readme.md https://huggingface.co/datasets/C3GroupCornell/RC-timescale-data/resolve/main/SimDataSystems/readme.md
822 Bytes
Directory Structure
/data/96IC_pkl: contains 96 trajectories for 8 different systems.
- dt = 0.001
- t_final = 1000
- System dimensions
- lorenz, nosehoover, sakaraya, and rossler are all 3 dimensional
- hyper lorenz, hyper xu, hyper bao, and hyper wang are 4 dimensional
model_weights/model_weights_pkl: contains randomly generated $W_r$ and $W_{in}$ matricies. Contains separate $W_{in}$ matricies for systems of dimension 3 and dimension 4
- Reservoir size $N_r = 400$
- $W_r$ density $\rho = 0.02$
- $W_r$ spectral radius $\lambda = 0.8$
- $W_{in}$ pulled from uniform $\mathcal{U}(-0.084, 0.084)$
notebooks/open_data_and_weights.ipynb: instructions to load in trajectories and model weights from pickle files
Other architecture choices:
- bias 1.6 (same for every layer)
- leak rate 0.6