Skip to content

downstream_models

Module defining the measurement maps from the encoded sequence to a single value prediction

Mave_Global_Epistasis_Measurement

source

Mave_Global_Epistasis_Measurement(
   model_params
)

Nonlinear GE Measurement process. for z_dim = 1 this is the same as in MAVE-NN

Args

  • model_params (dict) : Dictionary of model parameters
  • keys : num_hidden_nodes, z_dim

Attributes

  • num_hidden_nodes (int) : Number of hidden nodes in the measurement process.
  • z_dim (int) : Dimension of the latent space.

Methods:

.forward

source

.forward(
   z
)

Compute y_hat from sample of z

Linear_Measurement

source

Linear_Measurement(
   model_params
)

Linear Measurement head mapping z to yhat

Args

  • model_params (dict) : Dictionary of model parameters
  • keys : z_dim, mu_bias_init(optional)

Attributes

  • num_hidden_nodes (int) : Number of hidden nodes in the measurement process.
  • z_dim (int) : Dimension of the latent space.

Methods:

.forward

source

.forward(
   yh
)