If you like learning about AI, sign up for the [ Ссылка ]
In this video, I'm going to show you how to supercharge your GPT4All with the power of GPU activation. I'll also be using questions relating to hybrid cloud and edge computing for testing it.
If you're using GPT4All then you need to check out this video! By supercharging your GPT4All with the power of GPU activation, you'll be able to speed up you Q&A session and use the neweer LLaMA 2 model as well.
This is a video about GPT4All adding the power of the GPU.
I am not affiliated with the GPT4All Team or Nomic, I'm just an advocate for no-code Generative AI models being run on your own infrastructure.
The GPT4All website has tons of documentation, like getting to the GitHub repo to inspect the source code and details about using the API to use the model without the GUI.
[ Ссылка ]
[ Ссылка ]
If you are really excited by this project, you might also like to join their discord server: [ Ссылка ]
00:00 GPT4All Update to 2.4.15
00:52 Changes and choosing a model
02:19 Set Devices to 'Auto' for GPU Activation
02:44 Query the model
04:21 Watching the GPU and CPU Utilization
06:28 Falcon's Answer evaluation
08:00 Testing LLaMA 2 (and seeing the GPU activated)
10:45 Wrap-up
Several people have asked about the system config. Here are my system specs, complete with affiliate (see disclosure below paid) links.
I built this system to work from home, create and experience immersive graphics, write code, and mine over the Christmas holidays 2020, and I turned it on in Jan 2021. No regrets.
- Windows 10 Pro | [ Ссылка ]
- AMD Ryzen 9 3950X (16-core) | [ Ссылка ]
- ASUS TUF Gaming B550M-PLUS AMD AM4 (3rd Gen Ryzen™) Micro ATX Gaming Motherboard | [ Ссылка ]
- Noctua NH-D15S, Premium Dual-Tower CPU | [ Ссылка ]
- 64GB Ram (4x16GB) | [ Ссылка ]
- SAMSUNG 970 EVO SSD 1TB | [ Ссылка ]
- ASUS Dual NVIDIA GeForce RTX 3060 Ti OC Edition Gaming | [ Ссылка ]
- Montech Mid-Tower case | [ Ссылка ]
- Samsung Odyssey CRG Series 49" Curved Ultrawidescreen | [ Ссылка ]
If you are considering an upgrade, start with memory. Every day, I seem to have 10 more browser tabs open, and the website takes more and more all the time. If you want to build a new rig today and this "running local models" is a passion project for you, let me know in the comments if you are interested in a video discussing this in the future.
