Expand description
Transforms - Data Augmentation and Preprocessing
Provides composable transformations for data preprocessing and augmentation.
@version 0.1.0
@author AutomataNexus Development Team
Structs§
- Clamp
- Clamps tensor values to a specified range.
- Compose
- Composes multiple transforms into a single transform.
- Dropout
Transform - Applies dropout by randomly zeroing elements during training.
- Flatten
- Flattens the tensor to 1D.
- Lambda
- Applies a custom function as a transform.
- Normalize
- Normalizes a tensor with mean and standard deviation.
- Random
Crop - Randomly crops a portion of the input.
- Random
Flip - Randomly flips the input along a specified dimension.
- Random
Noise - Adds random Gaussian noise to the input.
- Reshape
- Reshapes the tensor to a specified shape.
- Scale
- Scales tensor values by a constant factor.
- ToTensor
- Converts input to a tensor (identity for already-tensor inputs).
Traits§
- Transform
- Trait for data transformations.