What if an LLM could use tools directly? In this post I discuss a potentially divergent view of agents, where agents are less like systems and more like LLMs specially trained to solve problems with a specific set of tools.
No pages have linked to this URL yet.