Your submission is now in Draft mode.

Once it's ready, please submit your draft for review by our team of Community Moderators. Thank you!

Submit Essay

Once you submit your essay, you can no longer edit it.

Pending

This content now needs to be approved by community moderators.

Submitted

This essay was submitted and is waiting for review.

When will there be a 50B parameter DALL-E?

Question

OpenAI's DALL-E was revealed on Jan 5 2021. It is a transformer trained on a multimodal dataset including text and images. It has 12 billion parameters.

The more famous GPT-3 has essentially the same architecture, but was trained only on text, and has 175 billion parameters.

When will credible sources reveal the existence of something similar to DALL-E but significantly bigger?

The date this question is interested in is the date the existence of the model becomes publicly known. Resolution by credible media report.

(As an aside, I'd love to make the question be about when the model is actually built, rather than when it is known of, but that's a lot harder to resolve because OpenAI typically doesn't reveal training completion dates.)

Here is some more detail on what "something similar to DALL-E but significantly bigger" means:

  • The model must be trained on images and text, though it can also be trained on other things

  • The model must be able to generate images given text descriptions, or images given text+images.

  • The model must have at least 50 billion parameters. Mixture-of-Experts models don't count; they instead should be thought of as a group of several smaller models.

This question closes retroactively the day before the model's existence first becomes known, in the sense described above.

Make a Prediction

Prediction

Note: this question resolved before its original close time. All of your predictions came after the resolution, so you did not gain (or lose) any points for it.

Note: this question resolved before its original close time. You earned points up until the question resolution, but not afterwards.

Current points depend on your prediction, the community's prediction, and the result. Your total earned points are averaged over the lifetime of the question, so predict early to get as many points as possible! See the FAQ.