Comment on From GPT-4 to GPT-5: Measuring progress through MedHELM [pdf]parentComments−manveerc1yIn the context of coding agents, they do simulate “reasoning” when you feed them the output and it is able to correct itself.
Comments
In the context of coding agents, they do simulate “reasoning” when you feed them the output and it is able to correct itself.