Get the latest tech news
OpenAI o1 Results on ARC-AGI-Pub
How far are the o1 preview and mini models from AGI?
Source: OpenAI " Learning to Reason with LLMs".In practice, o1 is significantly less likely to make mistakes when performing tasks where the sequence of intermediate steps is well-represented in the synthetic CoT training data. At training time, OpenAI says they've built a new reinforcement learning (RL) algorithm and a highly data-efficient process that leverages CoT. Automatic iterative re-prompting enables the model to better adapt to novelty, in a way similar to test-time fine-tuning leveraged by the MindsAI team.
Or read this on Hacker News
