-
Notifications
You must be signed in to change notification settings - Fork 205
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Lookup index for the trained models #22
Comments
I have an observation: take the last group of experiments on 3dshapes as an example, there are 1800 in total (10800~12599). As mentioned in the paper, they evaluated 6 methods, with 6 regularization strengths (i.e., hyperparameter values) each. And they also tried 50 different random As for the specific configurations, you may run some first and get the printed config as anchors, and calculate the index of your desired one. |
This is a late answer but I think it may be useful for other people looking for this kind of feature. For example for unsupervised_study_v1: from disentanglement_lib.config.unsupervised_study_v1.sweep import get_config
import pandas as pd
configs = get_config()
print(configs[12]) # give the config of model 12
df = pd.dataFrame(configs)
# Get the beta_tc_vae models using dsprites and a beta param of 4.
model = df['model.name'] == "beta_tc_vae"
dataset = df['dataset.name'] == "dsprites_full"
beta = df["beta_tc_vae.beta"] == 4.0
# Give all the possible model configs unused parameters excluded (e.g., annealed_vae.gamma)
print(df[model & dataset & beta].dropna(axis=1))
# Give the list of model ids with this configuration
print(df.index[model & dataset & beta]) |
Does a lookup index exist for the 10'800 pretrained disentanglement_lib modules? I wanted to quickly reproduce specific experiments (e.g. btcvae, beta=4, dsprites), but I am not sure how to find that pretrained model.
Thanks!
The text was updated successfully, but these errors were encountered: