News for hackers. Stuff that matters.

A comment on DALL·E: Introducing Outpainting. Parent comment

Re:DALL·E: Introducing Outpainting by singhrac on Wednesday August 31, 2022 @04:30PM (#32667945)
They’re just harder to run on your own resources, since large language models are very large. BLOOM was released a month ago, is likely better than GPT-3 in quality, and requires 8 A100s for inference, which pretty much no one has on their desk.
[ Parent | on HN ]
Re:DALL·E: Introducing Outpainting by visarga on Wednesday August 31, 2022 @04:36PM (#32668013)
Can anyone confirm if BLOOM is better than GPT-3 at instruction following? I might have read somewhere that it's not as well behaved.
[ Parent | on HN ]
Re:DALL·E: Introducing Outpainting by aljungberg on Wednesday August 31, 2022 @05:24PM (#32668559)
GPT-3 was fine-tuned after release to be better at following instructions. I don’t think that’s been done for BLOOM.

BLOOM incorporates some new ideas like ALiBi which might make it better in a more general sense. They haven’t released official evaluation numbers yet though so we’ll have to see.

[ Parent | on HN ]
Re:DALL·E: Introducing Outpainting by TakeBlaster16 on Wednesday August 31, 2022 @04:47PM (#32668150)
That makes sense, I didn't consider that angle. Thanks for the info.
[ Parent | on HN ]