Introducing FACTS Grounding. A new benchmark weβre launching with Google DeepMind to evaluate LLMβs factual accuracy on over 1700 tasks. π§ π
17.12.2024 15:36 β π 7 π 5 π¬ 1 π 1@natekeating.bsky.social
Naturally Foolish, Artificially Intelligent Head of Product @Kaggle
Introducing FACTS Grounding. A new benchmark weβre launching with Google DeepMind to evaluate LLMβs factual accuracy on over 1700 tasks. π§ π
17.12.2024 15:36 β π 7 π 5 π¬ 1 π 1And it was the Colab team that found the exploit and reported it in the GitHub issue you linked! github.com/googlecolab/...
Major credit to the team; nasty exploit that hit us too