SGREL2WebAppML

This webapp will fetch content and process it by certain model.

Model link: https://drive.google.com/drive/folders/1XelXeuxk3yqKF22bOAYPSUTURHhPHkwX

Usage:

Start WebApp:

python3 s2waml.py

---

Predict content:

POST /api/predict
with data: {'content': ..., 'model_name': ...}

Example result Model 1:

{"result":"CLAIM","score":0.9973957538604736}

Example result Model 2:

{"Claim":[1,0.9996739625930786],"Evidence":[0,0.06034946069121361],"Reasoning":[0,0.17518417537212372]}

---

Predict list of contents:

POST /api/predict
with data: {'content_list': [...], 'model_name': ...}

Example result:

{"result_list":[{"result":"CLAIM","score":0.9973957538604736},{...}]}

---

List model names:

GET /api/list_models

Example result:

{'default': 'model_1', 'models': [{'full_name': 'Some Model 1', 'name': 'model_1'}]}

---

Test all models:

python3 try_all_models.py

---

Install on Ubuntu 22.04:

Execute commands:

sudo apt update
sudo apt install python3-pip
sudo systemctl mask apt-daily.service
sudo systemctl disable apt-daily.timer
sudo systemctl disable unattended-upgrades
sudo systemctl disable --now apt-daily{,-upgrade}.{timer,service}
sudo shutdown -r now
sudo pip3 install sentencepiece torch flask flask_cors transformers requests
sudo nano /etc/systemd/system/sgrel2_ml_webapp.service

Write file contents:

[Unit]
After=network.target

[Service]
ExecStart=/usr/bin/python3 /home/ubuntu/SGREL2WebAppML/s2waml.py

[Install]
WantedBy=default.target

Execute commands:

sudo chmod 664 /etc/systemd/system/sgrel2_ml_webapp.service
sudo chmod 744 /home/ubuntu/SGREL2WebAppML/s2waml.py
sudo systemctl daemon-reload
sudo systemctl enable sgrel2_ml_webapp.service
sudo shutdown -r now

Set SGREL2 MySQL option api_settings -> 'ai_score_base_url' to new address