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
chenzk
bert_large_squad_onnxruntime
Commits
127d4fe2
Commit
127d4fe2
authored
Oct 10, 2023
by
chenxj
Browse files
update README
parent
ed679b4e
Changes
5
Hide whitespace changes
Inline
Side-by-side
Showing
5 changed files
with
24 additions
and
8 deletions
+24
-8
README.md
README.md
+21
-7
model.properties
model.properties
+3
-1
resources/bert_result.png
resources/bert_result.png
+0
-0
resources/squad.png
resources/squad.png
+0
-0
resources/transformer.png
resources/transformer.png
+0
-0
No files found.
README.md
View file @
127d4fe2
# bert_large_squad_onnx
# bert_large_squad_onnx
##
模型介绍
##
论文
bert-large的squad模型。
https://arxiv.org/pdf/1810.04805.pdf
## 模型结构
## 模型结构
基于transformer的结构
bert_large_squad核心是transformer,transformer结构如下:
## 推理
### 环境配置

## 算法原理
bert_large_squad模型的主要参数为:24个transformer层、1024个hidden size、16个self-attention heads,简要原理可用下图表示:

## 数据集
暂无合适中文数据集
## 环境配置
在
[
光源
](
https://www.sourcefind.cn/#/service-details
)
可拉取推理的docker镜像,在
[
光合开发者社区
](
https://cancon.hpccube.com:65024/4/main/
)
可下载onnxruntime安装包。bert_large_squad_onnx推荐的镜像如下:
在
[
光源
](
https://www.sourcefind.cn/#/service-details
)
可拉取推理的docker镜像,在
[
光合开发者社区
](
https://cancon.hpccube.com:65024/4/main/
)
可下载onnxruntime安装包。bert_large_squad_onnx推荐的镜像如下:
```
```
docker pull image.sourcefind.cn:5000/dcu/admin/base/custom:ort1.14.0_migraphx3.0.0-dtk22.10.1
docker pull image.sourcefind.cn:5000/dcu/admin/base/custom:ort1.14.0_migraphx3.0.0-dtk22.10.1
...
@@ -15,11 +22,13 @@ docker pull image.sourcefind.cn:5000/dcu/admin/base/custom:ort1.14.0_migraphx3.0
...
@@ -15,11 +22,13 @@ docker pull image.sourcefind.cn:5000/dcu/admin/base/custom:ort1.14.0_migraphx3.0
```
```
python3 fp16-convert.py
python3 fp16-convert.py
```
```
##
#
推理
## 推理
```
```
python3 main.py
python3 main.py
```
```
## 性能数据
## result

### 性能数据
fp32
fp32
| loop | time(ms) |
| loop | time(ms) |
...
@@ -49,6 +58,11 @@ fp16
...
@@ -49,6 +58,11 @@ fp16
| 9 | 0.048555332934483886 |
| 9 | 0.048555332934483886 |
| 10 | 0.048343464033678174 |
| 10 | 0.048343464033678174 |
## 应用场景
### 算法类别
nlp
### 热点应用行业
问答系统
## 源码仓库及问题反馈
## 源码仓库及问题反馈
https://developer.hpccube.com/codes/modelzoo/bert_large_squad_onnx
https://developer.hpccube.com/codes/modelzoo/bert_large_squad_onnx
## 参考
## 参考
...
...
model.properties
View file @
127d4fe2
# 模型唯一标识
modelCode
=
339
# 模型名称
# 模型名称
modelName
=
bert_large_squad_onnx
modelName
=
bert_large_squad_onnx
# 模型描述
# 模型描述
modelDescription
=
bert_large_squad_onnx是基于bert-large的squad downstream任务的onnx模型
modelDescription
=
bert_large_squad_onnx是基于bert-large的squad downstream任务的onnx模型
# 应用场景
# 应用场景
appScenario
=
推理,NLP
appScenario
=
推理,NLP
,问答
# 框架类型
# 框架类型
frameType
=
onnxruntime
frameType
=
onnxruntime
resources/bert_result.png
0 → 100644
View file @
127d4fe2
17.5 KB
resources/squad.png
0 → 100644
View file @
127d4fe2
78.4 KB
resources/transformer.png
0 → 100644
View file @
127d4fe2
150 KB
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