DALL-E icon indicating copy to clipboard operation
DALL-E copied to clipboard

Published 20 hours ago • verified publisher icon openai

Are the released model weights the best performing ones?

Open kracwarlock opened this issue 3 years ago • 4 comments

I found these checkpoints in the notebook:

  • https://cdn.openai.com/dall-e/encoder.pkl
  • https://cdn.openai.com/dall-e/decoder.pkl

Are these released model weights the same as the ones used for generating the blog post and paper images or are these a weaker version? I am asking because there has been a recent trend especially in NLP where people don't release their best models but slightly weaker ones.

kracwarlock avatar Apr 05 '21 17:04 kracwarlock

@adityaramesh

kracwarlock avatar Apr 16 '21 16:04 kracwarlock

any updates?

kracwarlock avatar Jul 20 '21 20:07 kracwarlock

Good question @kracwarlock! Also wondering how would you rehydrate these checkpoints ?

fosteman avatar Aug 16 '21 12:08 fosteman

I found these checkpoints in the notebook:

  • https://cdn.openai.com/dall-e/encoder.pkl
  • https://cdn.openai.com/dall-e/decoder.pkl

Are these released model weights the same as the ones used for generating the blog post and paper images or are these a weaker version? I am asking because there has been a recent trend especially in NLP where people don't release their best models but slightly weaker ones.

Is a new common practice

arqlz avatar Oct 07 '21 23:10 arqlz