ActiveTigger in Python
Project description
(py) Active Tigger
🚧 Under development 🚧
Server/client BERT fine tunning app using active learning
Python refactoring of R Shiny ActiveTigger app (Julien Boelaert & Etienne Ollion)
Installation
You can install activetigger
via pip (be careful to use Python 3.11):
pip install activetigger
Start the server
Create a config file config.yaml
in the directory where you want to launch the server :
path
: path to store files (for instance./data
)secret_key
: to secure tokens. To generate itopenssl rand -hex 32C
path_models
: absolute path to fasttext models
Then, to launch the server (on 0.0.0.0 port 8000 by default). You can configurate exposed port if needed.
python -m activetigger
Otherwise, you can launch only the API with uvicorn
:
uvicorn activetigger.api:app --host 0.0.0.0 --port 80 --reload
Users management
You can create/delete users with the root account.
You can also create a add_users.yaml
file with the users to add when the server starts :
user1: password1
user2: password2
Technical specifications
- REST-like client/server architecture
- Moslty stateless
- Mixed data storage : sqlite + files
- Backend Python
- FastAPI
- Multiprocessing for CPU-bound tasts (concurrent.futures.ProcessPoolExecutor)
- Frontend
- Streamlit (prototyping)
- React (in the future)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for activetigger-0.1.4-py2.py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | e944cce60cd44d837bdefb0e3e22856ea0d1f412d4e1e304263e1a3e4b96d87a |
|
MD5 | 5e1f0595b1ba87576fe4b37b8de5d0f4 |
|
BLAKE2b-256 | 8d2258e47349785ad1720eba3bc263af6adc27f7d2a700862394177e70605024 |