Hi everyone!
I am currently using the new Text-davinci-002 for a new tool. I have given some prompts for it to see how the output has to be created.
But I am receiving the same outputs in return. Here is a look at my playground. I have changed the Frequency and the presence penalty but they don’t make much of a difference. Please help me out a little.
Your instructions are not detailed enough. GPT-3 doesn’t know it’s a bot so it’s just trying to auto-complete but since the instructions are sparse it gets stuck on repeat. Also the “Human:” - “Bot:” thing makes it look like chat, which is probably also confusing GPT-3.
Thank you for your reply, Dave!
You mentioned that the “Human:” - “Bot:” thing makes it look like a chat, which makes sense. I want to make the prompts for a platform where I shall ask users about their inputs and in return GPT3 would give me the output.
Can you tell me another format wherein I can structure the prompts accordingly? I want to provide users with the ease of giving inputs and return them detailed outputs from GPT3.
Thanks
Just populate the above prompt from a web form.
Can you give me an example Dave? It would really help me.
