nntoolbox.tabular.models.collab module¶
-
class
nntoolbox.tabular.models.collab.CollabFiltering(n_users: int, n_products: int, embedding_dim: int)[source]¶ Bases:
torch.nn.modules.module.ModuleA simple collaborative model
-
forward(inputs: torch.Tensor) → torch.Tensor[source]¶ - Parameters
inputs – the pair of user-product of shape (batch_size, 2)
- Returns
(batch_size, 1)
-
get_score(users: torch.Tensor, products: torch.Tensor) → torch.Tensor[source]¶ Return the score of corresponding pairs of users-products
- Parameters
users – (batch_size, )
products – (batch_size, )
-
training: bool¶
-
-
class
nntoolbox.tabular.models.collab.NonLinearCF(n_users: int, n_products: int, user_dim: int, product_dim: int, body: Optional[torch.nn.modules.module.Module] = None)[source]¶ Bases:
torch.nn.modules.module.ModuleA non-linear collaborative model. If no body model is provided, default to a one-hidden layer net
-
forward(inputs: torch.Tensor) → torch.Tensor[source]¶ Return the score of corresponding pairs of products and users
- Parameters
inputs – the pair of user-product of shape (batch_size, 2)
- Returns
(batch_size, 1)
-
get_score(users: torch.Tensor, products: torch.Tensor) → torch.Tensor[source]¶ Return the score of corresponding pairs of users-products
- Parameters
users – (batch_size, )
products – (batch_size, )
-
training: bool¶
-