|
@@ -18,7 +18,7 @@ Run large language models with `llama.cpp`.
|
|
|
|
|
|
|
|
- Download for macOS
|
|
- Download for macOS
|
|
|
- Download for Windows (coming soon)
|
|
- Download for Windows (coming soon)
|
|
|
-- Docker: `docker run -p 8080:8080 ollama/ollama`
|
|
|
|
|
|
|
+- Docker: `docker run -p 11434:11434 ollama/ollama`
|
|
|
|
|
|
|
|
You can also build the [binary from source](#building).
|
|
You can also build the [binary from source](#building).
|
|
|
|
|
|
|
@@ -75,25 +75,19 @@ ollama run ~/Downloads/vicuna-7b-v1.3.ggmlv3.q4_1.bin
|
|
|
## Building
|
|
## Building
|
|
|
|
|
|
|
|
```
|
|
```
|
|
|
-
|
|
|
|
|
make
|
|
make
|
|
|
-
|
|
|
|
|
```
|
|
```
|
|
|
|
|
|
|
|
To run it start the server:
|
|
To run it start the server:
|
|
|
|
|
|
|
|
```
|
|
```
|
|
|
-
|
|
|
|
|
./ollama server &
|
|
./ollama server &
|
|
|
-
|
|
|
|
|
```
|
|
```
|
|
|
|
|
|
|
|
Finally, run a model!
|
|
Finally, run a model!
|
|
|
|
|
|
|
|
```
|
|
```
|
|
|
-
|
|
|
|
|
./ollama run ~/Downloads/vicuna-7b-v1.3.ggmlv3.q4_1.bin
|
|
./ollama run ~/Downloads/vicuna-7b-v1.3.ggmlv3.q4_1.bin
|
|
|
-
|
|
|
|
|
```
|
|
```
|
|
|
|
|
|
|
|
## API Reference
|
|
## API Reference
|
|
@@ -103,13 +97,7 @@ Finally, run a model!
|
|
|
Complete a prompt
|
|
Complete a prompt
|
|
|
|
|
|
|
|
```
|
|
```
|
|
|
-
|
|
|
|
|
curl --unix-socket ~/.ollama/ollama.sock http://localhost/api/generate \
|
|
curl --unix-socket ~/.ollama/ollama.sock http://localhost/api/generate \
|
|
|
-X POST \
|
|
-X POST \
|
|
|
-d '{"model": "/path/to/model", "prompt": "Once upon a time", "stream": true}'
|
|
-d '{"model": "/path/to/model", "prompt": "Once upon a time", "stream": true}'
|
|
|
-
|
|
|
|
|
-```
|
|
|
|
|
-
|
|
|
|
|
-```
|
|
|
|
|
-
|
|
|
|
|
```
|
|
```
|
Jeffrey Morgan