mirror of
https://github.com/open-webui/docs
synced 2025-05-21 03:35:55 +00:00
51 lines
1.0 KiB
Markdown
51 lines
1.0 KiB
Markdown
|
|
### 🛠️ Bring Your Own Ollama (BYO Ollama)
|
|
|
|
If Ollama is running on the **host machine** or another server on your network, follow these steps.
|
|
|
|
#### Step 1: Check Available Models
|
|
|
|
- If Ollama is **local**, run:
|
|
|
|
```bash
|
|
curl http://localhost:11434/v1/models
|
|
```
|
|
|
|
- If Ollama is **remote**, use:
|
|
|
|
```bash
|
|
curl http://<remote-ip>:11434/v1/models
|
|
```
|
|
|
|
This confirms that Ollama is available and lists the available models.
|
|
|
|
#### Step 2: Set the OLLAMA_BASE_URL
|
|
|
|
If Ollama is running **remotely** or on the host, set the following environment variable:
|
|
|
|
```bash
|
|
export OLLAMA_HOST=<remote-ip>:11434
|
|
```
|
|
|
|
This ensures Open WebUI can reach the remote Ollama instance.
|
|
|
|
#### Step 3: Download Llama 3.2
|
|
|
|
From your local or remote machine, run:
|
|
|
|
```bash
|
|
ollama pull llama3.2
|
|
```
|
|
|
|
**Tip:** Use this command to download the 8-bit version from Hugging Face:
|
|
|
|
```bash
|
|
ollama pull hf.co/bartowski/Llama-3.2-3B-Instruct-GGUF:Q8_0
|
|
```
|
|
|
|
#### Step 4: Access the WebUI
|
|
|
|
You can now access the WebUI at:
|
|
|
|
[http://localhost:3000](http://localhost:3000)
|