PADL

Pipeline Abstractions for Deep Learning

Full documentation here: https://lf1-io.github.io/padl/

Just as programs are read more often than they are written, so are deep learning models used more often than they are trained.

The PyTorch ecosystem has many tools for training models. However, this is only the beginning of the journey. Once a model has been trained, it will be shared, and used in a multitude of contexts, often on a daily basis, in operations, evaluation, comparision and experimentation by data scientists. The use of the trained model, is how value is extracted out of its weights. Despite this important fact, support for using deep-learning models up to now has been very thin in the PyTorch ecosystem and beyond. PADL is a tool which fills this void.

PADL

Is great for:

• serving model predictions
• experimentation with pre-trained models in Jupyter notebooks and interactive sessions
• inspecting and visualizing intermediate model features
• evaluating and monitoring model performance on metrics and incoming test data
• computing model outputs as a preprocessing step for further data science tasks

This is how it works:

• PADL is a pipeline builder for PyTorch.
• may be used with all of the great PyTorch functionality you're used to for writing layers.
• allows users to build pre-processing, forward passes, loss functions and post-processing into the pipeline.
• models may have arbitrary topologies and make use of arbitrary packages from the python ecosystem.
• allows for converting standard functions to PADL components using a single keyword transform.

PADL pipelines:

• can be trained with minimal boilerplate in PyTorch Lightning
• can then be served in one line with TorchServe
• interact well with the entire PyTorch ecosystem (for example Hugging Face)

PADL was developed at LF1, an AI innovation lab based in Berlin, Germany.

Getting Started

Installation

pip install padl

PADL currently supports python 3.7, 3.8 and 3.9.

Python version >= 3.8 is preferred because creating and loading transforms (not execution) can be slower in 3.7.

WARNING: PADL transforms do not work in the base Python Interpreter environment because we rely on the inspect module to find source code (used when saving PADL transforms). Unfortunately, the source code typed at this interactive prompt is discarded as soon as it is parsed. Therefore, we recommend using the IPython interpreter or Jupyter Notebooks for interactive sessions.

Your first PADL program

from padl import transform, batch, unbatch
import torch
from torch import nn
nn = transform(nn)

@transform
def prepare(x):
    return torch.tensor(x)

@transform
def post(x):
    return x.topk(1)[1].item()

my_pipeline = prepare >> batch >> nn.Linear(10, 20) >> unbatch >> post

Try out PADL in Colab notebooks

1. Basic PADL
2. MNIST
3. Simple NLP example
4. Sentiment Analysis - NLP
5. DC-GAN - Image Generation
6. CLIP guided diffusion for face editing

Resources

• Read the documentation at https://lf1-io.github.io/padl/.
• Find examples at https://github.com/lf1-io/padl/tree/main/notebooks.
• Post issues at https://github.com/lf1-io/padl/issues.

Dev Blog

Read more about PADL on the PADL developer's blog

Contributing

Code of conduct

If you're interested in contributing to PADL please look at the current issues

Licensing

PADL is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.