BaseImpl

class jlpy.pytorch.base.linear_regression.base_impl.BaseImpl

Bases: object

Deep Learning Linear Regression Base Implemetation.

1. Generate a demo training and validation dataset.

2. Build a linear regression model.

3. Train the LRModel.

   • Iterate over epoch.

   • Setup training mode.

     • Iterate over mini-batch of training set.

     • Optim.zero_grad.

     • Model forward.

     • Loss function.

     • Loss backward.

     • Optim step.

   • Setup validation mode.

     • Torch.no_grad.

     • Iterate over mini-batch of validation set.

     • Model forward.

     • Loss function.

4. Plot the result.

Methods

__init__	Construct a class instance.
fit	Train the model.
plot	Plot the loss.
show	Show the dataset.

__init__(*, bsize: int = 32, nepoch: int = 2, gap: int = 50, ws: list[float] = [2, -3.4], **kwargs: float) → None

Construct a class instance.

show(data: DataLoader[T], loop: int = 0) → None

Show the dataset.

Parameters:

• data (DataLoader[T]) – Dataset.

• loop (int) – Loops for iterate over dataset.

Returns:

None

Return type:

None

plot() → None

Plot the loss.

Returns:

None

Return type:

None

fit() → None

Train the model.

Returns:

None

Return type:

None