Those 4 steps: pre training, supervised fine tuning, reward model, policy is what exactly OpenAI did to ChatGPT That means when it replies, at the bottom there is thumbs up and down. I was thinking maybe that’s what human feedback is!
6 months ago | 0
Are you really appreciate all your content making easy complex issues with a very valuable neutrality. And your focus on open source is a great bet.
6 months ago | 0
IBM Technology
Hello IBM Technology subscribers and welcome to our Community Tab!
Thank you for being a valued member of our community here at IBM Technology. To help us make better content for you, please complete the following 3 min survey. We greatly appreciate your feedback to help us improve our channel!
your.feedback.ibm.com/jfe/form/SV_cXZtgDW9EqC6fwa
6 months ago | [YT] | 127