正在加载视频...

视频加载失败

DeepSeek-R1 on Mac Studio 192GB 🪄

259,545 次观看 • 1 年前 •via X (Twitter)

12 条评论

Georgi Gerganov 的头像
Georgi Gerganov1 年前

The model is quantized by @UnslothAI

Solar Heavy 的头像
Solar Heavy1 年前

- Moonlight

Unsloth AI 的头像
Unsloth AI1 年前

Looks soo good! 🎇🐋

Daniel Nakov 的头像
Daniel Nakov1 年前

come on apple, new studio

Georgi Gerganov 的头像
Georgi Gerganov1 年前

M4 studio with 256GB would be the final boss

unfazed. 的头像
unfazed.1 年前

@JudeSkrr il a encore drop

Vasilii Kharitonov 🌍 的头像
Vasilii Kharitonov 🌍1 年前

Tried R1 32B 8bit on M4 Pro with 48 GB memory (ggml-org/DeepSeek-R1-Distill-Qwen-32B-Q8_0-GGUF), seems to run great 👍 Memory usage is 43.3/48.0GB, eval time is about 7 tokens per second.

GP 的头像
GP1 年前

I need your cli setup 😏

Woody Lu 的头像
Woody Lu1 年前

Could you share the noise level (in DB) of the mac when running it locally?

Georgi Gerganov 的头像
Georgi Gerganov1 年前

Not sure how to measure it. It only makes noise for very long contexts, otherwise it's completely silent. And when it makes noise, it's not really loud IMO.

Alexander Kvamme 的头像
Alexander Kvamme1 年前

What profiling tool are you using? Much cleaner than htop

Georgi Gerganov 的头像
Georgi Gerganov1 年前

asitop

相关视频