Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
ModelZoo
NVIDIA-Nemotron-3_vllm
Commits
ebf6792b
Commit
ebf6792b
authored
Mar 20, 2026
by
raojy
💬
Browse files
Update README.md
parent
6d1d4299
Changes
1
Show whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
7 additions
and
7 deletions
+7
-7
README.md
README.md
+7
-7
No files found.
README.md
View file @
ebf6792b
...
...
@@ -23,7 +23,8 @@ Nemotron-3-Super-120B-A12B-BF16 是由英伟达 (NVIDIA) 训练的大语言模
| triton | 3.3.0+das.opt2.dtk2604.torch291.20260210.g1329924c |
| torch | 22.9.0+das.opt1.dtk2604.20260206.g275d08c2 |
| torch | 1.26.1 |
当前仅支持定制镜像:
`harbor.sourcefind.cn:5443/dcu/admin/base/custom:vllm0.15.1-ubuntu22.04-dtk26.04-0130-py3.10-20260220`
推荐使用镜像:
`harbor.sourcefind.cn:5443/dcu/admin/base/custom:vllm0.15.1-ubuntu22.04-dtk26.04-0130-py3.10-20260220`
-
挂载地址
`-v`
根据实际模型情况修改
...
...
@@ -32,7 +33,7 @@ Bash
```
docker run -it --shm-size 200g \
--network=host \
--name
<name>
\
--name
nemotron
\
--privileged \
--device=/dev/kfd \
--device=/dev/dri \
...
...
@@ -44,10 +45,7 @@ docker run -it --shm-size 200g \
-v /opt/hyhal/:/opt/hyhal/:ro \
harbor.sourcefind.cn:5443/dcu/admin/base/custom:vllm0.15.1-ubuntu22.04-dtk26.04-0130-py3.10-20260220 bash
```
关于本项目 DCU 显卡所需的特殊深度学习库,numpy、transformers 库需要替换安装:
Bash
关于本项目 DCU 显卡所需的特殊深度学习库,numpy、vllm 库需要替换安装:
```
pip uninstall vllm
...
...
@@ -72,7 +70,6 @@ pip install numpy==1.26.1
**注意**
:对于 120B 参数量的 BF16 模型,单机推理建议至少使用 8 张 K100 AI。使用时需添加
`--disable-custom-all-reduce`
参数。
Bash
```
## serve启动
...
...
@@ -111,6 +108,9 @@ curl http://localhost:8000/v1/chat/completions \
## 效果展示
<div
align=
center
>
<img
src=
"./doc/1.png"
/>
</div>
### 精度
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment