*https://www.youtube.com/watch?v=82TrKuAl7Ic
**
таймкоды
00:00:00 Введение
- Обсуждение предыдущих видео об OpenClaw и установке в изолированной виртуальной машине.
- Цель текущего видео: подключение OpenClaw к локальной модели ИИ на графическом процессоре.
00:00:39 Установка Ubuntu и подготовка
- Запуск терминала от имени администратора.
- Установка WSL и Ubuntu.
- Проверка наличия графического процессора NVIDIA.
- Установка ZSTD и Olama.
00:01:39 Загрузка и запуск модели
- Загрузка модели Olama с 7 миллиардами параметров.
- Запуск Olama с новой моделью.
- Настройка автоматического запуска Olama.
00:02:38 Установка OpenClaw и подключение модели
- Установка OpenClaw и выбор «Быстрый старт».
- Ввод данных для подключения Olama вручную.
- Получение URL панели управления с токеном.
00:03:36 Настройка локальной модели в OpenClaw
- Добавление записи для Olama в разделе «Провайдеры».
- Указание ключа API и базового URL.
- Настройка контекстных окон и ID модели.
00:04:03 Тестирование и выводы
- Тестирование доступа к локальной модели LLM.
- Обсуждение компромиссов по сравнению с OpenAI.
- Пример создания игры в змейку.
00:05:29 Заключение
- Призыв к зрителям попробовать метод и поделиться опытом в комментариях.
- Прощание и пожелание беречь себя.
In this video
Chapters
Transcript
Search transcript
Chapter 1: Introduction to OpenClaw
0:00
0 seconds
In our last couple of videos about OpenClaw, we covered installing it in a segmented VM for security and also locking down a WSL instance of OpenClaw.
0:09
9 seconds
However, both of those still relied on sending all of our prompts out to the cloud. Today, we’re going to take it a
0:16
16 seconds
step further and connect OpenClaw to a local AI model running directly on our GPU. For the demo, I’ll start from scratch, so feel free to follow along.
0:27
27 seconds
All of the commands can be found in the description for easy copy and paste reference.
Chapter 2: WSL and GPU Configuration
0:39
39 seconds
Crack open a terminal and be sure to run it as admin. If you don’t already have WSL installed, go ahead and run this
0:46
46 seconds
command to spin up Ubuntu. I will be fast forwarding through the installs, but they only take a few minutes. Once the installer completes, you’ll be
0:54
54 seconds
prompted to create a username and password. Make sure you don’t forget or lose this information. Now that Ubuntu
1:01
1 minute, 1 second
is installed, we need to make sure we can see our GPU. If you have an Nvidia graphics card, it should show up right away. Here we see my RTX 3080 with 10
1:11
1 minute, 11 seconds
gigs of VRAM. Next, we need to grab ZSTD in preparation for the following step.
1:18
1 minute, 18 seconds
Drop that command in now. Okay, now that we have that, we can install Olama. Pop in that command. Now, this will take a
Chapter 3: Installing Ollama and Models
1:26
1 minute, 26 seconds
few minutes to complete. All right, now that OAMA is installed, let’s pull down a model. I’m going with a 7B model as
1:34
1 minute, 34 seconds
that’s about what I can load into VRAMm on my 380. Drop that command. Now, if you want a different model, just make note of it as we’ll need that info later
1:42
1 minute, 42 seconds
on when plugging it into Open Claw. Once that finishes, let’s run Olama with the new model. Drop that command in. Now,
1:52
1 minute, 52 seconds
this takes us to a prompt where we can chat with the model. I’ll ask it a random question. Watch how fast this is.
1:59
1 minute, 59 seconds
Literally instant. Don’t get too excited, though. It’s not that fast when we plug it into OpenClaw. Type in forward slashby to get out of the
2:07
2 minutes, 7 seconds
prompt. Our next command will set to start up automatically. Pop that in now.
Chapter 4: OpenClaw Installation
2:14
2 minutes, 14 seconds
All right, we are all set. And now it’s time to install OpenClaw. Drop that command in now. If prompted for your password, plug that in and let it cook.
2:24
2 minutes, 24 seconds
This will take a few minutes to complete and then take us straight into the onboarding process. Acknowledge that this is powerful and therefore risky.
2:34
2 minutes, 34 seconds
Select quick start. For model, select skip for now. Select all providers, then choose to enter manually. Delete the
2:43
2 minutes, 43 seconds
default entry and type in slash [music] our model name. We will skip setting up any channels, skills, and hooks for now.
2:53
2 minutes, 53 seconds
Give it a sec and the gateway will install and start up. [music] Scroll up to the dashboard URL with token. Copy
3:02
3 minutes, 2 seconds
that and pop it into a browser. We don’t have any model set up yet, so OpenClaw won’t respond if we try chatting with
3:09
3 minutes, 9 seconds
it. We need to configure our local model. Now, on the left side, click on config. Scroll down and click on models.
Chapter 5: Configuring Local Model
3:18
3 minutes, 18 seconds
Click on all. Scroll down to the mode section and click on add entry under
3:24
3 minutes, 24 seconds
providers. We will name it O Lama. For API, drop that down and select Olama.
3:33
3 minutes, 33 seconds
For API key, this is important. Olama doesn’t use any keys, but open claw will be expecting [music] one. Just pop in
3:42
3 minutes, 42 seconds
any arbitrary value here, but don’t leave it blank. For off, select API key.
3:50
3 minutes, 50 seconds
For base URL, enter http colon slash slash127.0.0.1 col11434.
4:03
4 minutes, 3 seconds
This is the base URL for Olama’s API.
4:07
4 minutes, 7 seconds
Now, scroll down to models and click add. Scroll down to context windows and enter 200,000.
4:15
4 minutes, 15 seconds
Now scroll down to ID and enter our exact model name. For max tokens, pop in [music] 8,192.
4:26
4 minutes, 26 seconds
Then give it a name. I use the model name here as well. All right, go ahead and save that. And then we can head back to chat for testing. Now we let it know
4:35
4 minutes, 35 seconds
our name and give it a name as well as a vibe and some general info. And voila, we have a working open claw with a local
Chapter 6: Testing and Limitations
4:43
4 minutes, 43 seconds
LLM. No data leaving our machine unless we choose. So now, I would be lying if I told you that this will work as well as
4:52
4 minutes, 52 seconds
hooking it up to OpenAI or another paid LLM. But hey, there are trade-offs. If privacy is at the top of your list, this
5:00
5 minutes
might be an ideal setup for you. One thing I noticed is that this particular model had a hard time actually writing
5:08
5 minutes, 8 seconds
to my file system compared to using OpenAI where it just works pretty much every time. After some wrestling with
5:15
5 minutes, 15 seconds
it, I was able to get it to create a basic little snake game for me. Just an example of the capabilities.
5:29
5 minutes, 29 seconds
Now that brings back memories [music] just as I recall back in 10th grade English class. Not too shabby for
5:37
5 minutes, 37 seconds
running this straight from my little GPU. Let us know in the comments if you will be trying this out and if so, what
5:43
5 minutes, 43 seconds
GPU are you going to use? I hope you all enjoyed the video. Until the next one, take care [music] everyone.
5:59
5 minutes, 59 seconds
[music]
6:04
6 minutes, 4 seconds
[music]
6:06
6 minutes, 6 seconds
dude.
Sync to video time

