3 ms·
Very interesting! Can you see those helping for RAG scenarios? Specifically: - decreasing models tendency to answer with ungrounded answers - increase models
by pamelafox 3y ago
Very interesting! Can you see those helping for RAG scenarios? Specifically:
- decreasing models tendency to answer with ungrounded answers
- increase models ability to respond with the correct syntax for citations- the open models like llama2 dont seem to obey my prompt’s syntax instructions.
- batch12 3y agoFor the second item, I've had luck using grammar to overcome this issue. The easiest one to implement I've seen so far is Microsoft's guidance-ai.
- sanxiyn 3y agoYou can use outlines https://github.com/outlines-dev/outlines https://github.com/outlines-dev/outlines to let models generate with correct syntax.
- pamelafox 3y agoThanks! I havent had to use a syntax-enforcing framework with gpt-35, I’ll try outlines and guidance out to see if they help enforce syntax for the locally runnable models.