TensorLy is a Python library that aims at making tensor learning simple and accessible. It allows to easily perform tensor decomposition, tensor learning and tensor algebra. Its backend system allows to seamlessly perform computation with NumPy, MXNet or PyTorch and run methods at scale on CPU or GPU.
- Website: http:https://tensorly.org
- Source-code: https://github.com/tensorly/tensorly
- Jupyter Notebooks: https://github.com/JeanKossaifi/tensorly-notebooks
The only pre-requisite is to have Python 3 installed. The easiest way is via the Anaconda distribution.
With pip (recommended) | With conda |
pip install -U tensorly
|
conda install -c tensorly tensorly
|
Development (from git) | |
# clone the repository
git clone https://github.com/tensorly/tensorly
cd tensorly
# Install in editable mode with `-e` or, equivalently, `--editable`
pip install -e .
|
Note: TensorLy depends on NumPy by default. If you want to use the MXNet or PyTorch backends, you will need to install these packages separately.
For detailed instruction, checkout the documentation.
Testing and documentation are an essential part of this package and all functions come with uni-tests and documentation.
The tests are ran using the pytest package (though you can also use nose). First install pytest:
pip install pytest
Then to run the test, simply run, in the terminal:
pytest -v tensorly
Alternatively, you can specify for which backend you wish to run the tests:
TENSORLY_BACKEND='numpy' pytest -v tensorly
import tensorly as tl
import numpy as np
Create a small third order tensor of size 3 x 4 x 2 and perform simple operations on it:
tensor = tl.tensor(np.arange(24).reshape((3, 4, 2)))
tl.unfolded = unfold(tensor, mode=0)
tl.fold(unfolded, mode=0, shape=tensor.shape)
Applying tensor decomposition is easy:
from tensorly.decomposition import tucker
# Apply Tucker decomposition
core, factors = tucker(tensor, rank=[2, 2, 2])
# Reconstruct the full tensor from the decomposed form
tl.tucker_to_tensor(core, factors)
Changing the backend to perform computation on GPU for instance. Note that using MXNet or PyTorch requires to have installed them first:
tl.set_backend('pytorch') # Or 'mxnet' or 'numpy'
Now all the computation is done by PyTorch, and tensors can be created on GPU:
import torch
tensor = tl.tensor(np.arange(24).reshape((3, 4, 2)), dtype=torch.cuda.FloatTensor)
type(tensor) # torch.cuda.FloatTensor
For more information on getting started, checkout the user-guide and for a detailed reference of the functions and their documentation, refer to the API
If you see a bug, open an issue, or better yet, a pull-request!