manyoso/haltt4llm

This project is an attempt to create a common metric to test LLM's for progress in eliminating hallucinations which is the most serious current problem in widespread adoption of LLM's for many real purposes.

36 best points · 16 comments · 2 Hacker News posts · 223 GitHub stars · View on GitHub

Hacker News discussions

• GPT4All outscores GPT-3.5 on new hallucination metric (36 points, 13 comments, by manyoso, 2023-03-30)
• New metric for LLM hallucinations with results for GPT-3, GPT3-5 and Alpaca Lora (4 points, 3 comments, by manyoso, 2023-03-29)