Whats the advantage of RL here? Instead of using the better foundational models with a thinking tool and the typical ReACT pattern etc? I guess token costs & latency are the main advantages? Does that make it worth it? (Genuine question, interested to understand the why behind doing this)
Hey! I saw your post pop up on my homepage and wanted to show some support. If you get a chance, I’d really appreciate a little love on my latest newsletter too always happy to boost each other!
Whats the advantage of RL here? Instead of using the better foundational models with a thinking tool and the typical ReACT pattern etc? I guess token costs & latency are the main advantages? Does that make it worth it? (Genuine question, interested to understand the why behind doing this)
Hey! I saw your post pop up on my homepage and wanted to show some support. If you get a chance, I’d really appreciate a little love on my latest newsletter too always happy to boost each other!