mirror of
https://github.com/deepseek-ai/deepseek-recipe.git
synced 2026-10-28 17:17:08 +08:00
93b0fd229c
Co-authored-by: xinchengxx <xinchengxx@deepseek.com> Co-authored-by: zhu-he <hezhu@deepseek.com>
47 lines
1.5 KiB
Markdown
47 lines
1.5 KiB
Markdown
# server-rs
|
|
|
|
An Axum API example using the Rust components with mock inference. The
|
|
request's model name selects the DeepSeek V4 or V4.1 prompt encoding:
|
|
`DeepSeek-V4-Pro` and `deepseek-v4-pro` select V4, `DeepSeek-V4.1-Flash` and
|
|
`deepseek-flash` select V4.1, and conversion rejects any other name with a
|
|
`400` error. Images use OpenCV preprocessing.
|
|
|
|
## Usage
|
|
|
|
Install the
|
|
[Rust and OpenCV prerequisites](../docs/development.md#native-dependencies),
|
|
then run from the repository root:
|
|
|
|
```sh
|
|
cargo run -p server-rs --locked
|
|
```
|
|
|
|
The server listens on `127.0.0.1:7777`. Set `SERVER_ADDR` to change the address.
|
|
|
|
## API
|
|
|
|
- `POST /v1/chat/completions`
|
|
- `POST /v1/responses`
|
|
- `POST /v1/messages`
|
|
|
|
All three protocols support JSON for `stream=false` and SSE for `stream=true`.
|
|
|
|
For example, in a second terminal:
|
|
|
|
```sh
|
|
curl http://127.0.0.1:7777/v1/chat/completions \
|
|
-H 'Content-Type: application/json' \
|
|
-d '{"model":"deepseek-flash","messages":[{"role":"user","content":"Hello"}],"stream":false}'
|
|
```
|
|
|
|
The response contains the mock answer `Hello world!`. Set `"stream":true` and
|
|
add `-N` to curl to inspect SSE output. No API key or model weights are
|
|
required.
|
|
|
|
## Deployment scope
|
|
|
|
This example has no authentication and uses an image fetcher that does not
|
|
filter private network addresses. Keep it on loopback for local use; see the
|
|
[deployment considerations](../docs/development.md#example-servers) before
|
|
exposing it to other users. The Python example uses the same default port.
|