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
Qwen3-ASR_pytorch
Commits
9c3935eb
Commit
9c3935eb
authored
Apr 30, 2026
by
weishb
Browse files
更新readme版本
parent
3959ccbd
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
9 additions
and
10 deletions
+9
-10
README.md
README.md
+9
-10
No files found.
README.md
View file @
9c3935eb
...
@@ -48,6 +48,13 @@ pip install numpy==1.25.0
...
@@ -48,6 +48,13 @@ pip install numpy==1.25.0
pip
install
-r
requirements.txt
pip
install
-r
requirements.txt
```
```
## 预训练权重
**请根据`支持的DCU型号`选择对应模型下载,FP8模型仅在BW1100/BW1101上支持,其他型号请勿使用!**
| 模型名称 | 权重大小 | 数据类型 | 支持的DCU型号 | 最低卡数需求 | 下载地址 |
|:-----:|:----------:|:----------:|:----------:|:---------------------:|:----------:|
| Qwen3-ASR-1.7B | 1.7B | BF16 | BW1000 | 1 |
[
ModelScope
](
https://www.modelscope.cn/models/Qwen/Qwen3-ASR-1.7B
)
|
| Qwen3-ASR-0.6B | 0.6B | BF16 | BW1000 | 1 |
[
ModelScope
](
https://www.modelscope.cn/models/Qwen/Qwen3-ASR-0.6B
)
|
## 数据集
## 数据集
暂无
暂无
...
@@ -65,14 +72,14 @@ python inference.py
...
@@ -65,14 +72,14 @@ python inference.py
### vllm
### vllm
#### 单机推理
#### 单机推理
```
bash
```
bash
#
# serve启动
# serve启动
export
VLLM_USE_FUSED_RMS_ROPE
=
0
export
VLLM_USE_FUSED_RMS_ROPE
=
0
vllm serve Qwen/Qwen3-ASR-1.7B
\
vllm serve Qwen/Qwen3-ASR-1.7B
\
--trust-remote-code
\
--trust-remote-code
\
--limit-mm-per-prompt
'{"audio": 1}'
--limit-mm-per-prompt
'{"audio": 1}'
#
# client访问
# client访问
curl
-X
POST
"http://127.0.0.1:8000/v1/audio/transcriptions"
\
curl
-X
POST
"http://127.0.0.1:8000/v1/audio/transcriptions"
\
-F
"file=@/path/to/test.wav"
\
-F
"file=@/path/to/test.wav"
\
-F
"model=Qwen/Qwen3-ASR-1.7B"
-F
"model=Qwen/Qwen3-ASR-1.7B"
...
@@ -88,14 +95,6 @@ curl -X POST "http://127.0.0.1:8000/v1/audio/transcriptions" \
...
@@ -88,14 +95,6 @@ curl -X POST "http://127.0.0.1:8000/v1/audio/transcriptions" \
### 精度
### 精度
`DCU与GPU精度一致,推理框架:transformers、vllm`
`DCU与GPU精度一致,推理框架:transformers、vllm`
## 预训练权重
**请根据`支持的DCU型号`选择对应模型下载,FP8模型仅在BW1100/BW1101上支持,其他型号请勿使用!**
| 模型名称 | 权重大小 | 数据类型 | 支持的DCU型号 | 最低卡数需求 | 下载地址 |
|:-----:|:----------:|:----------:|:----------:|:---------------------:|:----------:|
| Qwen3-ASR-1.7B | 1.7B | BF16 | BW1000 | 1 |
[
Modelscope
](
https://www.modelscope.cn/models/Qwen/Qwen3-ASR-1.7B
)
|
| Qwen3-ASR-0.6B | 0.6B | BF16 | BW1000 | 1 |
[
Modelscope
](
https://www.modelscope.cn/models/Qwen/Qwen3-ASR-0.6B
)
|
## 源码仓库及问题反馈
## 源码仓库及问题反馈
-
https://developer.sourcefind.cn/codes/modelzoo/qwen3-asr_pytorch
-
https://developer.sourcefind.cn/codes/modelzoo/qwen3-asr_pytorch
...
...
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