Very difficult to get the attention of decision makers. More joint marketing opportunities for worthy initiatives. Incredible GPT-3 based projects not receiving the support they should. C-suite execs not accessible/too remote. Well meaning job worth's stymying innovation.

because "the API is not available in your country."

Most definitely the lack of token weights within an example when fine-tuning. This would open up a lot of new possibilities. We currently have to train our own models to circumvent this limitation

Would love to see DALL-E in the API. Don’t have much of a use case for text gen right now, but there’s a project for image gen that I have been eager to implement for months now

Fine-tune model for classification precision/recall was almost equal to self hosted BERTish model. So we keep using self hoted one.

As an OpenAI GPT-3 user and Dall-E 2 user, I have no complaints and I can't think of anything that could be better. I freaking love OpenAI and hope it will keep on existing for many more decades to come! 😀😀😀😀😀😀

This is coming from a developer point of view... I just don't find it that useful. Though its cool but again i don't have a need for these tools. :/

the uncertainty about codex pricing. buidling products without knowing what the api will cost some day makes pricing calculations difficult (especially worried since dalle2 is way more expensive than expected)

I think better token details within the billing/admin details. When generating lots of data it's hard to pin point which things have been chewing up the budget

Some way of reasoning about larger projects with multiple files would be good