add deepseek-v3 amd docker command (#3495)
This commit is contained in:
@@ -98,3 +98,21 @@ drun sglang_image \
|
|||||||
```
|
```
|
||||||
|
|
||||||
With your AMD system properly configured and SGLang installed, you can now fully leverage AMD hardware to power SGLang’s machine learning capabilities.
|
With your AMD system properly configured and SGLang installed, you can now fully leverage AMD hardware to power SGLang’s machine learning capabilities.
|
||||||
|
|
||||||
|
## Running DeepSeek-V3
|
||||||
|
|
||||||
|
The only difference in running DeepSeek-V3 is when starting the server.
|
||||||
|
|
||||||
|
```bash
|
||||||
|
drun -p 30000:30000 \
|
||||||
|
-v ~/.cache/huggingface:/root/.cache/huggingface \
|
||||||
|
--ipc=host \
|
||||||
|
--env "HF_TOKEN=<secret>" \
|
||||||
|
sglang_image \
|
||||||
|
python3 -m sglang.launch_server \
|
||||||
|
--model deepseek-ai/DeepSeek-V3 # <- here \
|
||||||
|
--tp 8 \
|
||||||
|
--trust-remote-code \
|
||||||
|
--host 0.0.0.0 \
|
||||||
|
--port 30000
|
||||||
|
```
|
||||||
Reference in New Issue
Block a user