@sama the incidence of API outages over the last three months is a little concerning. What can be done to guarantee high availability of the APIs?

Rate limiting by far is a pain. But thankfully OpenAI’s customer support is A+

@sama with GPT3 for commercial applications, its important to ensure a certain level of consistency in the response with linguistic styling. A feature to ensure a certain style of response is returned/not returned would be great.

Amazing product! 2 things: - Way to reassure corp customers: how data is used by you + way to deploy on premise/their servers - Guarantees you will not verticalize me using my fine tuning + prompt + all the ideas I had All this while allowing you to retain what is owed to you ofc

1. Unlimited history with Dall E. I can’t find my older creations 2. Dall E bulk creation: I want run multiple prompts in parallel. If I have 5 ideas I dislike doing them 1 by 1 3. Dall E auto complete - Suggest modifications to my query

Fine tune model loading is too slow API costs are extremely expensive

DallE's erase+fill is hampered by needing to describe the whole image. When result is poor, not clear if because of problems parsing overall prompt, or the small alteration. Would be better to only describe the missing part. Upload photo of room, erase wall, say"floating shelves"

I've been playing with the Edit option in Dalle, I use a chrome plug in to resize the images directly on the website and re run the prompt. I would like that option to be native, also for those situations the AI should let you rework with a realistic face already created there.

It is about Helion energy. Add our technology and will get fusion in 2024. have a look profile