MiDuRy-SentiBal

Sentiment analysis on generative AI tweets.

Dataset

Generative AI Tweets — Kaggle

---

Setup (run once)

1. Clone the repo

git clone https://github.com/jonesrmj/MiDuRy-SentiBal.git
cd MiDuRy-SentiBal

2. Create a virtual environment and install dependencies

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

3. Open the notebook in VS Code

• Open MiDuRy-SentiBal.ipynb
• Top right → Select Kernel → Python Environments → select venv
• Run all cells

---

Daily workflow

git pull                  # get latest changes
source venv/bin/activate  # activate venv

Then open the notebook in VS Code and run.

Google Colab Notes

Use Colab for training only, not editing. Make edits locally → push to GitHub → pull in Colab to run. If you edit on Colab, save back to GitHub immediately before anyone else pushes.

Colab does not auto-save your trained model. The runtime is wiped on disconnect. Save the model to Google Drive after training:

from google.colab import drive
drive.mount('/content/drive')
model.save_pretrained('/content/drive/MyDrive/MiDuRy-SentiBal/model')

Note: RoBERTa is ~500MB — too large for GitHub, use Drive instead.

---

Repo Structure

data/
  tweets.csv              # dataset (included in repo)
MiDuRy-SentiBal.ipynb    # main notebook
requirements.txt          # dependencies