3 ms·

Can OpenAI's GPT-OSS Beat a 10-Year-Old?

• ticulatedspline 1y ago
  > "According to the experiment results, the AI successfully navigated many of the questions posed in the test, showcasing its ability to mimic certain aspects of human-like understanding and problem-solving." Not sure the author actually read that source, it doesn't paint a rosy picture of it's performance. 9 out of 80 at default and "better" but still bad with a 32K context https://www.windowscentral.com/artificial-intelligence/i-asked-openais-new-open-source-llm-to-complete-a-test-designed-for-children-is-it-smarter-than-a-10-year-old https://www.windowscentral.com/artificial-intelligence/i-ask...