I write quite a few models, for practice, work, research or fun. Most of these models are pretty small. Their size simply does not warrant the creation of another repo, hence all such models will be uploaded, here. I plan to upload the model files as well. Every transformers based model will be hosted on huggingface spaces for inference and an open api which anyone can use. My goal with this project is to encourage myself to work on more utility based models and integrating them as this task should be extremely easy using the api.
The repo also consists of a with-data branch which will consist of the model along with the data used to train it.
Each model also comes with a gradio frontend.
All the models are present in their own folder along with a model state file of some kindpreq. It may be pickle (.pkl) or pytorch(.pt or .pth). Each fpreqtxt file. This must be used to install all dependencies. I highly recommend using a virtual environment for this.
To create a virtual environment and use it
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
If you want to use the model locally, it can be used with gradio
cd <name_of_model>
python app.py
For inference, all transformers models/apis can be accessed from my Hugging Face Account