Zero-registration, local AI toolkit for offline text summarization, question-answering, audio transcription, and image captioning on Debian/Ubuntu derivatives.
- Text summarization via HuggingFace T5
- Question-answering with DistilBERT-SQuAD
- Audio transcription using OpenAI Whisper (local)
- Image captioning with Salesforce BLIP
ai_toolkit/
├── requirements.txt # Python dependencies
├── app.py # CLI entrypoint
└── tasks/
├── summarizer.py # summarize(text)
├── qa.py # answer(question, context)
├── transcription.py # transcribe(audio_path)
└── image_caption.py # caption_image(image_path)
sudo apt update
sudo apt install -y python3-venv ffmpeg git
git clone https://github.com/youruser/ai_toolkit.git
cd ai_toolkit
python3 -m venv venv && source venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt# Summarize a text file:
python app.py summarize path/to/article.txt
# Question-Answering on a document:
python app.py qa "What is the main topic?" path/to/article.txt
# Transcribe an audio file:
python app.py transcribe path/to/audio.mp3
# Generate a caption for an image:
python app.py caption path/to/photo.jpgEach command will print the result to stdout.
- Wrap with Flask or FastAPI for a web UI
- Deploy models on GPUs by installing
torchwith CUDA support - Schedule tasks via
cronor systemd timers
- Fork the repo
- Create a branch (
git checkout -b feat/awesome) - Commit changes (
git commit -m "feat: add X") - Push (
git push origin feat/awesome) - Open a Pull Request
Please follow the Code of Conduct.