2 MacBooks are all you need with @ac_crypto @exolabs_ mo_baioumy & @ronaldmannak

Опубликовано: 24 Май 2025
на канале: Ray Fernando

999

Livestream with Alex Cheema M to talk about 2 MacBooks are all you need.

@ac_crypto @exolabs_ mo_baioumy & @ronaldmannak

🕒 Key moments:
00:00 Live stream intro
00:06 Two MacBooks run large language model
00:31 Distributing inference across multiple computers
01:16 Open source model beats closed-source models
02:00 Downloading 800GB model weights
02:58 Demo: Inference on two laptops
05:19 Project backstory
08:55 Audio issues resolved, show restart
10:24 Key accomplishment: Open-source distributed software
13:06 Explanation of inference
16:45 Apple Silicon's advantage in running large models
22:40 Comparison of cloud vs. local AI inference
29:50 Addressing audience question about CPU usage
30:43 Explanation of Exo's architecture
36:54 Future of AI: Accessibility and doomsday scenarios
41:14 Discussion about inference cost-effectiveness on CPU vs. GPU
48:55 Breakdown of events leading to successful Lama 3.1 demo
54:45 Open source contributions and hiring
57:15 Example applications for EXO
01:00:00 Live demo using Apollo app
01:12:00 Discussion on scaling and improving XO performance
01:32:50 Breakthroughs and future optimizations for XO
02:00:00 Edge device discussion