Slowly, then suddenly!

👑 of 🇺🇸 Open-source!

@0xSero nvidia open-sourcing models sells more gpus because every finetune off them runs best on cuda

@0xSero And let’s not forget about @arcee_ai 🚀

@0xSero Nvidia really released a lot last week.

@0xSero Но у Nvidia не всегда опенсорс реальных моделей, часто это proof-of-work или proof-of-concept.

А все остальные модели — опять китайцы или тюны китайских моделей. =)

@0xSero nvidia said forget the gpu money we’re just gonna flood huggingface too. at this point just rename it nvidiahugface and call it a day

@0xSero Pretty clear that whoever gets to be the “Linux” (as in most used open-source) of AI will win big.

Surprised it’s taking US companies so long to realize this.

@0xSero Microsoft AI open releases was what no one predicted

@0xSero That's really good we need more!🔥🔥🔥

@0xSero 老黄是真拼啊😂

@0xSero and magenta is so damn fun, absolutely worth messing around with. takes 2 minutes to set up and runs on consumer hardware.

@0xSero Brother

the representation of Asian American employees in the NVIDIA workforce was 49.2% which increased 2 percentage points from 2021

@0xSero Nvidia so chill fr they just want u to use there chips

@natolambert Please check out the fully open sourced robot learning benchmark I posted!

@natolambert 9/30 is wild, nvidia owns the whole stack now

@natolambert In reality, the most of people use Qwen3.6 models 😜

@0xSero hello world

@0xSero None of them are traditional text LLM’s

@natolambert LLMs follow this pattern exactly