Samplers¶

Here are implemented the main samplers used in the pythae.models.

By convention, each implemented model is contained within a folder located in pythae.samplers and named likewise the sampler. The following modules can be found in this folder:

samplername_config.py: Contains a SamplerNameConfig instance inheriting from BaseSamplerConfig where the sampler configuration is stored and
samplername_sampler.py: An implementation of the sampler_name inheriting from BaseSampler.
samplername_utils.py (optional): A module where utils methods are stored.

`BaseSampler`	Base class for samplers used to generate from the VAEs models.
`NormalSampler`	Samples from a Standard normal distribution in the Autoencoder’s latent space.
`GaussianMixtureSampler`	Fits a Gaussian Mixture in the Autoencoder’s latent space.
`TwoStageVAESampler`	Fits a second VAE in the Autoencoder’s latent space.
`HypersphereUniformSampler`	Sampling from uniform distribution on hypersphere.
`PoincareDiskSampler`	Sampling from the Poincaré Disk using either a Wrapped Riemannian or Riemannian Gaussian distribution.
`VAMPSampler`	Sampling from the VAMP prior.
`RHVAESampler`	Sampling form the inverse of the metric volume element of a `RHVAE` model.
`MAFSampler`	Fits a Masked Autoregressive Flow in the Autoencoder’s latent space.
`IAFSampler`	Fits an Inverse Autoregressive Flow in the Autoencoder’s latent space.
`PixelCNNSampler`	Fits a PixelCNN in the VQVAE’s latent space.

Basic Examples¶

To launch the data generation process from a trained model, you only need to build your sampler. For instance, to generate new data with your sampler, run the following.

Normal sampling¶

>>> from pythae.models import VAE
>>> from pythae.samplers import NormalSampler
>>> # Retrieve the trained model
>>> my_trained_vae = VAE.load_from_folder(
...     'path/to/your/trained/model'
... )
>>> # Define your sampler
>>> my_samper = NormalSampler(
...     model=my_trained_vae
... )
>>> # Generate samples
>>> gen_data = my_samper.sample(
...     num_samples=50,
...     batch_size=10,
...     output_dir=None,
...     return_gen=True
... )

Gaussian mixture sampling¶

>>> from pythae.models import VAE
>>> from pythae.samplers import GaussianMixtureSampler, GaussianMixtureSamplerConfig
>>> # Retrieve the trained model
>>> my_trained_vae = VAE.load_from_folder(
...     'path/to/your/trained/model'
... )
>>> # Define your sampler
...     gmm_sampler_config = GaussianMixtureSamplerConfig(
...     n_components=10
... )
>>> my_samper = GaussianMixtureSampler(
...     sampler_config=gmm_sampler_config,
...     model=my_trained_vae
... )
>>> # fit the sampler
>>> gmm_sampler.fit(train_dataset)
>>> # Generate samples
>>> gen_data = my_samper.sample(
...     num_samples=50,
...     batch_size=10,
...     output_dir=None,
...     return_gen=True
... )