The first framework that can post train DeepSeek V4-pro on a single-node?

3 pointsby timxzz2개월 전7 comments

Hi all,

We just opensourced a project called Orbit, which can RL post train trillion scale LLMs like deepseek v4. We found it pretty cool! Looking forward to some comments!

- Tim

Blog post:

https://spherelab.ai/orbit/

7 comments

freakynit2개월 전

This is so good. Is it kinda similar in concept to QLoRA training?

zqiu2개월 전

Hi, thank you so much :) Yes, you are very sharp, to make truely it scalable, we have to finetune quantized LLMs directly, similar ideas as QLoRA. The algorithmic innovation is based on our research about orthogonal finetuning OFT and QOFT, which has demonstrated better training stability when finetuning quantized base models in such a scale.

freakynit2개월 전

Thank you for the information, creating this and releasing the research behind it. Truly appreciate.

kendy19922개월 전

Looks very cool!!!

zqiu2개월 전

Thank you so much, glad you liked it :)

crookedusage902개월 전

wow impressive

zqiu2개월 전

Thank you so much :)