News for hackers. Stuff that matters.
Re:Building the Future of TensorFlow by reckless on Friday October 21, 2022 @08:45AM (#33287411)
I feel like the comments on backwards compatibility are due to the absolute shitshow of TF2 compatibility for TF1 code and models.

Also the threat of Pytorch can be seen when reading between the lines, especially since it's now run by a foundation and the darling of the diffusion model developments.

Re:Building the Future of TensorFlow by learndeeply on Friday October 21, 2022 @02:14PM (#33291228)
PyTorch has been a darling of almost every noteworthy open source model for the past 3-4 years (BLOOM, GPT-J, StyleGAN3, detectron, etc). Personally, I've only seen people use TensorFlow/XLA if they got free TPU credits from Google (gpt-neo), or if it was released by Google (t5).
Re:Building the Future of TensorFlow by hustwindmaple1 on Friday October 21, 2022 @07:29PM (#33294053)
GPT-J is actually JAX
Re:Building the Future of TensorFlow by learndeeply on Friday October 21, 2022 @10:59PM (#33295284)
My mistake, I was thinking of gpt-neox. Thanks for the correction.