Fluence 0.1.4

Pytorch based deep learning library focussed on providing computationally efficient low resource methods and algorithms

Fluence

Fluence is a Pytorch based deep learning library focussed on providing computationally efficient, low resource methods and algorithms. Although the main focus is to provide support with transformers, it can be extended with other architectures as well.

Installing

python3 setup.py install --user

The library contains implementation for the following approaches (many more to come):

• Adaptive Methods

  • Adaptive Attention Span in Transformers
  • Adaptively Sparse Transformers
  • Reducing Transformer Depth on Demand with Structured Dropout

• Meta Learning

• Optimizers:

  • Lamb
  • Lookahead

• Importance Sampling:

  • Clustering

Documentation

Please head to this link to learn how you can integrate fluence with your workflow. Since it's an early release, there might be bugs here and there. Please file an issue if you encounter one.

Tests

Tests can be run with pytest

pytest tests/

Author: Prajjwal Bhargava (@prajjwal_1)