[ comments | sourced from HackerNews ]
[HN] A GPT-4 capability forecasting challenge
Submitted 1 year ago by irradiated@radiation.party [bot] to technews@radiation.party
https://nicholas.carlini.com/writing/llm-forecast/question/Capital-of-Paris
S13Ni@lemmy.studio 1 year ago
This was really good! I self host stable diffusion and LLMs and have used chat gpt at work and still only got 50% right.