I just ordered 1000 Jev models nex week, byebye luna

[jerryzhang]

中帝卷哥们，快，我看好你们哦

[JianguoChuan]

Anthropic and OpenAi have not response to Jev yet. I think they have to update their models to do the same thing as Jev. But Jev took 2 years to do it, how long do they need to catch up?

[jerryzhang]

This shit just hit the fan! training on cpu 6hous, Training! fuck, i believe it can run on smartphone ...

[jerryzhang]

this is terra level, I imagine next week will be a chinese model beat astra like shit

[JianguoChuan]

this is terra level, I imagine next week will be a chinese model beat astra like shit

It is also good for US models, because they can further make the models larger and larger to use all GPUs.

[jerryzhang]

I see the opposite, 1million models, each for one purpose…run on cpu...nvda dead

[jerryzhang]

It is also good for US models, because they can further make the models larger and larger to use all GPUs.

没仔细研究它模型细节，但是以前学rl的时候记得 这东西在gpu跑不快 跟cpu差不多...

[jerryzhang]

baby r u ready, here we go...first one come, runs entirely on a CPU with 1–2 GB of memory (I haven't spent much time optimizing it yet), responds in 25–300 ms, and beats JEV in all benchmarks.