vatrix 0.1.1

Log Processor & SBERT Training Tool

🧠 Vatrix

Vatrix is a command-line tool for processing log data, rendering natural language descriptions using Jinja2 templates, injecting augmented data, and generating training pairs for fine-tuning Sentence Transformers (SBERT).

---

✨ Features

• CLI-powered processing of logs
• Modular template system powered by Jinja2
• SBERT data generation and similarity scoring
• Supports file mode, stream mode, and CLI flags
• Exports training pairs to CSV
• Exports highly similar sentence pairs for SBERT fine-tuning
• Flexible and colorful logging with log rotation

---

📦 Installation

pip install vatrix

Or install the latest from source:

git clone https://github.com/brianbatesactual/vatrix.git
cd vatrix
make setup

---

🛠️ Usage

vatrix --mode file \
       --render-mode all \
       --input data/input_logs.json \
       --output data/processed_logs.csv \
       --unmatched data/unmatched_logs.json \
       --generate-sbert-data \
       --log-level DEBUG \
       --log-file logs/vatrix_debug.log

Makefile Commands

make setup         # Create venv and install dependencies
make run           # Run log processor on default file
make stream        # Start reading NDJSON from stdin
make retrain       # Export SBERT sentence pairs
make freeze        # Regenerate requirements.txt
make clean         # Clean environment and build artifacts
make nuke          # Full reset of the project environment

---

🧠 Example

---

🧪 Testing

make test

---

📁 Logs

All logs are saved to the logs/ directory with daily rotation.

---

🧼 Cleanup

make clean    # Clean temp data
make nuke     # Wipe and rebuild virtualenv

---

📚 License

MIT © Brian Bates

Built with ❤️ for log intelligibility and NLP adventures.