Vibe Coding Discovervegaml
View on GitHubTyped decisions from a frozen language model and a small physics engine: calibrated probabilities, conformal sets, 73k context, images.
Python library for typed choices, scores, and boolean decisions using local frozen language models. It provides calibrated probabilities, conformal answer sets, abstention, image input, and support for 73k-token contexts.
Use Cases
Email and support-ticket routingCustomer churn predictionIntent classificationDocument classification from text or imagesSentiment and emotion classificationLong-document decision extraction
Built With
- Language
- Python
- Frameworks
- PyTorch · Transformers · Hugging Face Hub
Tags
typed decisions · local inference · classification · calibrated probabilities · conformal prediction · abstention · long context · multimodal · fine-tuning · frozen language model