Thank you. Can you give me any documents for this or any videosUse the latest chatpt model 4 turbo model. Check the documentation to find out.
If you are using command chaining, meaning one command after another. There are 2 ways to do it.
1) on the second command send the output of the first command so that you retain the context through out the command chaining. Do the same for third, fourth command and so on. This is crude but easy way of doing it.
2) you tell the openai then all the subsequent commands have the same context and are part of command chain. You do this via api on the latest models. It's little complicated but doable.
But there is little performance degradation, try to give better command and try to correct the same mistakes it's doing accoss multiple outputs.
I use bubble.io to create apps and make openai api calls. There are plenty of youtube vids about it. Check openai api docs here : platform.openai.com/docs/api-referenceThank you. Can you give me any documents for this or any videos
Does it use for PHP as well?I use bubble.io to create apps and make openai api calls. There are plenty of youtube vids about it. Check openai api docs here : platform.openai.com/docs/api-reference
Its a no-code platform so there is no need of code most of the time and no php. You can add javascript for advanced code.Does it use for PHP as well?
My tool has been made using php, that's why asking if it will help to refineIts a no-code platform so there is no need of code most of the time and no php. You can add javascript for advanced code.
I have made a tool to scale up content creation. When I check my prompt on chatgpt 3.5 and 4, it does give perfect output.
But when i put the prompt on my tool by using openai API, it doesn’t give me the correct outcome.
Why this happens?
You are so correct. I have seen 0.8 temp output is good.Check your temperature and top_p parameter. These affect the output quality dramatically. Low temperature produces more predictable a highly guided but stupid texts, while high temperature produce more creative content by may start they can become delusional and generate false information.
My suggestion is to try all your GPT prompts here, playing with the parameters: https://platform.openai.com/playground?mode=chat