check_shapes
is a library for annotating and checking tensor shapes.
For example:
import tensorflow as tf
from gpflow.experimental.check_shapes import check_shapes
@tf.function
@check_shapes(
"features: [batch..., n_features]",
"weights: [n_features]",
"return: [batch...]",
)
def linear_model(features: tf.Tensor, weights: tf.Tensor) -> tf.Tensor:
return tf.einsum("...i,i -> ...", features, weights)
For more information see our documentation.
The recommended way to install check_shapes
is from pypi:
pip install check_shapes
To develop check_shapes
, check it out from GitHub:
git clone [email protected]:GPflow/check_shapes.git
We use Poetry to install and manage dependencies. Follow their instructions for how to install Poetry itself. Then:
cd check_shapes
poetry install
To check you installation run our tests:
poetry run task test
For testing with different versions of Python and dependencies, see the poetryenv
script.