arcjax
129 posts


Oh well it seems @googlecloud is shutting down the TRC program 😢
So many memories and I learn so much from it. Sadly. Maybe if we will make ourself hear they gona continue supporting us Opensource guys? :)) I hope so!

English

@HeydariAI @test_tm7873 @googlecloud did you get this after asking for an extension, or is this the message telling you to get off it
English

@HeydariAI @googlecloud You too?! 😢my 3 other freidns got the same message. so sad
English

@test_tm7873 @googlecloud i know a guy who has his till 2027, i wonder what will happen then!
English

@arcjax7 @googlecloud All my 3 friends got the same message. 😿 So I was thinking it's the end. But honestly sadq they over the limit. Sad they thrown out old V3/V2s atleast that would help with the lot of users.
English

So software built with @zml_ai runs transparently at max speed on:
- CPU
- NVIDIA GPUs
- AMD GPUs
- Google TPUs
- AWS Trainium
- Intel GPUs
- Tenstorrent NPUs
- Apple GPUs (very experimental)
And more to come.
Steeve Morin@steeve
*Tensor parallel* multi-GPU in 2x Intel Arc B70s now runs at SOTA speed
English

The poaching by Microsoft to AllenAI was huge, over half of the authors!
And I can think of a few other ones not in this paper
Gabriele Berton@gabriberton
Cool paper by AllenAI and CMU on how to pretrain LLMs for long-context understanding The setup is legit and the findings are interesting, and they show that most LLMs have architectural issues 🧵
English

*Tensor parallel* multi-GPU in 2x Intel Arc B70s now runs at SOTA speed

Steeve Morin@steeve
Intel oneAPI support for @zml_ai PR now open at github.com/zml/zml/pull/5… Multi-GPU support doesn't work yet, but single GPU runs at SOTA. This now the 5th architecture we support, and 2 more are incoming.
English











