dpfm_factory

dpfm_factory is a Python package that provides a factory function to easily load different machine learning models and their associated preprocessing pipelines from Hugging Face. This package is particularly useful in the digital and computational pathology domains, where it is crucial to work with various specialized models.

Features

Easy Model Loading: Load machine learning models and their preprocessing pipelines with a simple factory function.
Hugging Face Integration: Seamlessly integrates with Hugging Face to authenticate and load models.
Custom Environment Variables: Supports loading environment variables from a .env file for sensitive data like tokens.

Installation

To install the package directly from GitHub, use the following command:

pip install git+https://github.com/Steven-N-Hart/dpfm_factory

Ensure that you have all necessary dependencies listed in the requirements.txt file. Alternatively, clone the repository and install the package locally:

git clone https://github.com/Steven-N-Hart/dpfm_factory
cd dpfm_factory
pip install -r requirements.txt
pip install .

Usage

Setup

Before using the package, make sure to create a .env file in the root of your project directory with your Hugging Face token:

HUGGINGFACE_TOKEN=your_huggingface_token_here

Example Usage

Here’s an example of how to use the model_factory function to load a model and its associated processor:

from dpfm_factory.model_runners import model_factory

# Specify the model you want to load
model_name = 'MahmoodLab/conch'

# Load the model, processor, and the function to get image embeddings
model, processor, get_image_embedding = model_factory(model_name)

# Example usage with an image (replace 'your_image' with actual image data)
image_embedding = get_image_embedding(your_image)

print("Image Embedding:", image_embedding)

Supported Models

The model_factory function currently supports the following models:

Note: Some of these models require a HuggingFace Token that has been approved by the model owner. The Google model also requires submitters to sign additional Terms of Service.

Error Handling

If an unsupported model name is provided, the model_factory will raise a NotImplementedError. For example:

try:
    model, processor, get_image_embedding = model_factory('unsupported/model_name')
except NotImplementedError as e:
    print(e)

Contributing

Contributions are welcome! Please fork the repository and submit a pull request with your changes.

Release files for dpfm-factory 0.13.7

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for dpfm-factory 0.13.7
File	Size	Uploaded
dpfm_factory-0.13.7.tar.gz	27.9 MB	Apr 1, 2025	Details

Built distribution (wheel)

Table of built distributions (wheels) for dpfm-factory 0.13.7
File	Interpreter	ABI	Platform	Reset
dpfm_factory-0.13.7-py3-none-any.whl 27.9 MB Apr 1, 2025	Python 3	none	any	Details

Total release size: 55.8 MB

Release files / dpfm_factory-0.13.7.tar.gz

Download URL	dpfm_factory-0.13.7.tar.gz
Size	27.9 MB
Tags	Source
SHA-256 checksum How to use checksums	`fe28da017132e37ffaf8b4c9245089e928966a68a983e443c2e8bf2cecf8b8ee`
BLAKE2b-256 checksum How to use checksums	`ef16015dafb001fc4b01949d7a15e3e79149f48208c01fd4b50c10ec334599c1`
Upload date	Apr 1, 2025
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/6.1.0 CPython/3.10.12`

Release files / dpfm_factory-0.13.7-py3-none-any.whl

Download URL	dpfm_factory-0.13.7-py3-none-any.whl
Size	27.9 MB
Tags	Python 3
SHA-256 checksum How to use checksums	`6b5f23252a47b034b2f9b9ba12e426360c6f0919a6be32baa6ab8158b3be30a9`
BLAKE2b-256 checksum How to use checksums	`d3d594d38468f8140165e7035473567631a61d8116c9ced01a97e84358f5184e`
Upload date	Apr 1, 2025
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`twine/6.1.0 CPython/3.10.12`