Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[Doc] Fixed quantize.md #795

Merged
merged 39 commits into from
Dec 5, 2022
Merged

[Doc] Fixed quantize.md #795

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
39 commits
Select commit Hold shift + click to select a range
f1e7bff
add onnx_ort_runtime demo
wjj19950828 Nov 11, 2022
25fdeb0
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 11, 2022
b5dff8d
rm in requirements
wjj19950828 Nov 11, 2022
5566e88
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 14, 2022
4e1f35a
support batch eval
wjj19950828 Nov 14, 2022
87dba08
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 15, 2022
99c610c
fixed MattingResults bug
wjj19950828 Nov 15, 2022
c59fd5c
move assignment for DetectionResult
wjj19950828 Nov 15, 2022
f82486d
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 15, 2022
ec475d0
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 16, 2022
7f68106
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 16, 2022
8bbea4f
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 16, 2022
b0f7c90
Merge remote-tracking branch 'upstream/develop' into update_runtime
wjj19950828 Nov 17, 2022
d428d95
Merge remote-tracking branch 'upstream/develop' into add_x2paddle
wjj19950828 Nov 17, 2022
0371811
integrated x2paddle
wjj19950828 Nov 18, 2022
c16eb42
Merge remote-tracking branch 'upstream/develop' into add_x2paddle
wjj19950828 Nov 18, 2022
279f160
Merge remote-tracking branch 'upstream/develop' into add_x2paddle
wjj19950828 Nov 20, 2022
cb9c966
add model convert readme
wjj19950828 Nov 20, 2022
010c0b3
Merge remote-tracking branch 'upstream/develop' into add_x2paddle
wjj19950828 Nov 20, 2022
670cefb
update readme
wjj19950828 Nov 20, 2022
6e9e6a5
re-lint
wjj19950828 Nov 20, 2022
c76f960
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Nov 22, 2022
2996743
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Nov 22, 2022
b0f45f7
add processor api
wjj19950828 Nov 22, 2022
6f797da
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Nov 22, 2022
5725868
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Nov 24, 2022
05a198b
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Nov 25, 2022
b85cc13
Add MattingResult Free
wjj19950828 Nov 25, 2022
1cfa9a6
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Nov 28, 2022
52180e7
change valid_cpu_backends order
wjj19950828 Nov 28, 2022
a71c403
Merge branch 'develop' into add_api_doc
jiangjiajun Nov 28, 2022
273e778
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Dec 1, 2022
43c9567
add ppocr benchmark
wjj19950828 Dec 1, 2022
f666166
Merge branch 'add_api_doc' of https://github.com/wjj19950828/FastDepl…
wjj19950828 Dec 1, 2022
ccfe6ff
mv bs from 64 to 32
wjj19950828 Dec 2, 2022
cb5436b
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Dec 2, 2022
90bd93e
Merge remote-tracking branch 'upstream/develop' into add_api_doc
wjj19950828 Dec 5, 2022
910bc4a
fixed quantize.md
wjj19950828 Dec 5, 2022
8d8362b
fixed quantize bugs
wjj19950828 Dec 5, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 5 additions & 5 deletions docs/cn/quantize.md
100644 → 100755
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,7 @@ FastDeploy基于PaddleSlim的Auto Compression Toolkit(ACT), 给用户提供了
目前, FastDeploy支持自动化压缩,并完成部署测试的模型的Runtime Benchmark和端到端Benchmark如下所示.

Benchmark表格说明:
- Rtuntime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- Runtime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- 端到端时延为模型在实际推理场景中的时延, 包含模型的前后处理.
- 所测时延均为推理1000次后求得的平均值, 单位是毫秒.
- INT8 + FP16 为在推理INT8量化模型的同时, 给Runtime 开启FP16推理选项
Expand All @@ -63,7 +63,7 @@ Benchmark表格说明:
| [YOLOv7](../../examples/vision/detection/yolov7/quantize/) | Paddle Inference | CPU | 995.85 | 477.93|None|None | 2.08 |51.1 | 46.2|量化蒸馏训练 |

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [YOLOv5s](../../examples/vision/detection/yolov5/quantize/) | TensorRT | GPU | 24.61 | 21.20 | 20.78 | 20.94 | 1.18 | 37.6 | 36.7 | 量化蒸馏训练 |
| [YOLOv5s](../../examples/vision/detection/yolov5/quantize/) | Paddle-TensorRT | GPU | 23.53 | None | 21.98 | 19.84 | 1.28 | 37.6 | 36.8 | 量化蒸馏训练 |
Expand Down Expand Up @@ -94,7 +94,7 @@ Benchmark表格说明:
| [MobileNetV1_ssld](../../examples/vision/classification/paddleclas/quantize/) | Paddle Inference | CPU | 12.29 | 4.68 | None|None|2.62 |77.89 | 71.36 |离线量化 |

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 Top1 | INT8 Top1 | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 Top1 | INT8 Top1 | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [ResNet50_vd](../../examples/vision/classification/paddleclas/quantize/) | TensorRT | GPU | 4.92| 2.28|2.24|2.23 | 2.21 | 79.12 | 79.06 | 离线量化 |
| [ResNet50_vd](../../examples/vision/classification/paddleclas/quantize/) | Paddle-TensorRT | GPU | 4.48|None |2.09|2.10 | 2.14 | 79.12 | 79.06 | 离线量化 |
Expand All @@ -119,7 +119,7 @@ NOTE:
- TensorRT比Paddle-TensorRT快的原因是在runtime移除了multiclass_nms3算子

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [ppyoloe_crn_l_300e_coco](../../examples/vision/detection/paddledetection/quantize ) | TensorRT | GPU | 35.75 | 15.42 |20.70|20.85 | 2.32 | 51.4 | 50.7 | 量化蒸馏训练 |
| [ppyoloe_crn_l_300e_coco](../../examples/vision/detection/paddledetection/quantize ) | Paddle-TensorRT | GPU | 33.48 |None | 18.47 |18.03 | 1.81 | 51.4 | 50.5| 量化蒸馏训练 |
Expand All @@ -134,6 +134,6 @@ NOTE:
| [PP-LiteSeg-T(STDC1)-cityscapes](../../examples/vision/segmentation/paddleseg/quantize) | Paddle Inference | CPU | 1138.04| 602.62 |None|None | 1.89 |77.37 | 71.62 |量化蒸馏训练 |

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mIoU | INT8 mIoU | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mIoU | INT8 mIoU | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [PP-LiteSeg-T(STDC1)-cityscapes](../../examples/vision/segmentation/paddleseg/quantize) | Paddle Inference | CPU | 4726.65| 4134.91|None|None | 1.14 |77.37 | 71.62 |量化蒸馏训练 |
4 changes: 2 additions & 2 deletions examples/vision/classification/paddleclas/quantize/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ FastDeploy 提供了一键模型自动化压缩工具, 能够简单地通过输
用户也可以直接下载下表中的量化模型进行部署.

Benchmark表格说明:
- Rtuntime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- Runtime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- 端到端时延为模型在实际推理场景中的时延, 包含模型的前后处理.
- 所测时延均为推理1000次后求得的平均值, 单位是毫秒.
- INT8 + FP16 为在推理INT8量化模型的同时, 给Runtime 开启FP16推理选项
Expand All @@ -33,7 +33,7 @@ Benchmark表格说明:
| [MobileNetV1_ssld](https://bj.bcebos.com/paddlehub/fastdeploy/mobilenetv1_ssld_ptq.tar) | Paddle Inference | CPU | 12.29 | 4.68 | None|None|2.62 |77.89 | 71.36 |离线量化 |

### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 Top1 | INT8 Top1 | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 Top1 | INT8 Top1 | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [ResNet50_vd](https://bj.bcebos.com/paddlehub/fastdeploy/resnet50_vd_ptq.tar) | TensorRT | GPU | 4.92| 2.28|2.24|2.23 | 2.21 | 79.12 | 79.06 | 离线量化 |
| [ResNet50_vd](https://bj.bcebos.com/paddlehub/fastdeploy/resnet50_vd_ptq.tar) | Paddle-TensorRT | GPU | 4.48|None |2.09|2.10 | 2.14 | 79.12 | 79.06 | 离线量化 |
Expand Down
4 changes: 2 additions & 2 deletions examples/vision/detection/paddledetection/quantize/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ FastDeploy 提供了一键模型自动化压缩工具, 能够简单地通过输


Benchmark表格说明:
- Rtuntime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- Runtime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- 端到端时延为模型在实际推理场景中的时延, 包含模型的前后处理.
- 所测时延均为推理1000次后求得的平均值, 单位是毫秒.
- INT8 + FP16 为在推理INT8量化模型的同时, 给Runtime 开启FP16推理选项
Expand All @@ -32,7 +32,7 @@ NOTE:
- TensorRT比Paddle-TensorRT快的原因是在runtime移除了multiclass_nms3算子

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [ppyoloe_crn_l_300e_coco](https://bj.bcebos.com/paddlehub/fastdeploy/ppyoloe_crn_l_300e_coco_qat.tar ) | TensorRT | GPU | 35.75 | 15.42 |20.70|20.85 | 2.32 | 51.4 | 50.7 | 量化蒸馏训练 |
| [ppyoloe_crn_l_300e_coco](https://bj.bcebos.com/paddlehub/fastdeploy/ppyoloe_crn_l_300e_coco_qat.tar ) | Paddle-TensorRT | GPU | 33.48 |None | 18.47 |18.03 | 1.81 | 51.4 | 50.5 | 量化蒸馏训练 |
Expand Down
4 changes: 2 additions & 2 deletions examples/vision/detection/yolov5/quantize/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,7 @@ FastDeploy 提供了一键模型自动化压缩工具, 能够简单地通过输
用户也可以直接下载下表中的量化模型进行部署.(点击模型名字即可下载)

Benchmark表格说明:
- Rtuntime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- Runtime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- 端到端时延为模型在实际推理场景中的时延, 包含模型的前后处理.
- 所测时延均为推理1000次后求得的平均值, 单位是毫秒.
- INT8 + FP16 为在推理INT8量化模型的同时, 给Runtime 开启FP16推理选项
Expand All @@ -29,7 +29,7 @@ Benchmark表格说明:
| [YOLOv5s](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5s_quant.tar) | Paddle Inference| CPU | 213.73 | 130.19 | None | None | 1.64 |37.6 | 35.2 | 量化蒸馏训练 |

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [YOLOv5s](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5s_quant.tar) | TensorRT | GPU | 24.61 | 21.20 | 20.78 | 20.94 | 1.18 | 37.6 | 36.7 | 量化蒸馏训练 |
| [YOLOv5s](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5s_quant.tar) | Paddle-TensorRT | GPU | 23.53 | None | 21.98 | 19.84 | 1.28 | 37.6 | 36.8 | 量化蒸馏训练 |
Expand Down
4 changes: 2 additions & 2 deletions examples/vision/detection/yolov6/quantize/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,7 +9,7 @@ FastDeploy 提供了一键模型自动化压缩工具, 能够简单地通过输
用户也可以直接下载下表中的量化模型进行部署.(点击模型名字即可下载)

Benchmark表格说明:
- Rtuntime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- Runtime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- 端到端时延为模型在实际推理场景中的时延, 包含模型的前后处理.
- 所测时延均为推理1000次后求得的平均值, 单位是毫秒.
- INT8 + FP16 为在推理INT8量化模型的同时, 给Runtime 开启FP16推理选项
Expand All @@ -28,7 +28,7 @@ Benchmark表格说明:


#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [YOLOv6s](https://bj.bcebos.com/paddlehub/fastdeploy/yolov6s_ptq_model.tar) | TensorRT | GPU | 15.66 | 11.30 | 10.25 |9.59 | 1.63 | 42.5 | 40.7|量化蒸馏训练 |
| [YOLOv6s](https://bj.bcebos.com/paddlehub/fastdeploy/yolov6s_ptq_model.tar) | Paddle-TensorRT | GPU | 15.03 | None| 11.36 | 9.32 | 1.61 | 42.5 | 40.7|量化蒸馏训练 |
Expand Down
4 changes: 2 additions & 2 deletions examples/vision/detection/yolov7/quantize/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ FastDeploy 提供了一键模型自动化压缩工具, 能够简单地通过输


Benchmark表格说明:
- Rtuntime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- Runtime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- 端到端时延为模型在实际推理场景中的时延, 包含模型的前后处理.
- 所测时延均为推理1000次后求得的平均值, 单位是毫秒.
- INT8 + FP16 为在推理INT8量化模型的同时, 给Runtime 开启FP16推理选项
Expand All @@ -29,7 +29,7 @@ Benchmark表格说明:
| [YOLOv7](https://bj.bcebos.com/paddlehub/fastdeploy/yolov7_quant.tar) | Paddle Inference | CPU | 995.85 | 477.93|None|None | 2.08 |51.1 | 46.2|量化蒸馏训练 |

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mAP | INT8 mAP | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [YOLOv7](https://bj.bcebos.com/paddlehub/fastdeploy/yolov7_quant.tar) | TensorRT | GPU | 36.47 | 18.81 | 20.33| 17.58| 2.07 | 51.1| 50.4|量化蒸馏训练 |
| [YOLOv7](https://bj.bcebos.com/paddlehub/fastdeploy/yolov7_quant.tar) | Paddle-TensorRT | GPU | 37.06|None|20.26|17.53 | 2.11 | 51.1| 50.4|量化蒸馏训练 |
Expand Down
4 changes: 2 additions & 2 deletions examples/vision/segmentation/paddleseg/quantize/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,7 +11,7 @@ FastDeploy 提供了一键模型自动化压缩工具, 能够简单地通过输
用户也可以直接下载下表中的量化模型进行部署.(点击模型名字即可下载)

Benchmark表格说明:
- Rtuntime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- Runtime时延为模型在各种Runtime上的推理时延,包含CPU->GPU数据拷贝,GPU推理,GPU->CPU数据拷贝时间. 不包含模型各自的前后处理时间.
- 端到端时延为模型在实际推理场景中的时延, 包含模型的前后处理.
- 所测时延均为推理1000次后求得的平均值, 单位是毫秒.
- INT8 + FP16 为在推理INT8量化模型的同时, 给Runtime 开启FP16推理选项
Expand All @@ -26,7 +26,7 @@ Benchmark表格说明:
| [PP-LiteSeg-T(STDC1)-cityscapes](https://bj.bcebos.com/paddlehub/fastdeploy/PP_LiteSeg_T_STDC1_cityscapes_without_argmax_infer_QAT_new.tar)) | Paddle Inference | CPU | 1138.04| 602.62 |None|None | 1.89 |77.37 | 71.62 |量化蒸馏训练 |

#### 端到端 Benchmark
| 模型 |推理后端 |部署硬件 | FP32 Runtime时延 | INT8 Runtime时延 | INT8 + FP16 Runtime时延 | INT8+FP16+PM Runtime时延 | 最大加速比 | FP32 mIoU | INT8 mIoU | 量化方式 |
| 模型 |推理后端 |部署硬件 | FP32 End2End时延 | INT8 End2End时延 | INT8 + FP16 End2End时延 | INT8+FP16+PM End2End时延 | 最大加速比 | FP32 mIoU | INT8 mIoU | 量化方式 |
| ------------------- | -----------------|-----------| -------- |-------- |-------- | --------- |-------- |----- |----- |----- |
| [PP-LiteSeg-T(STDC1)-cityscapes](https://bj.bcebos.com/paddlehub/fastdeploy/PP_LiteSeg_T_STDC1_cityscapes_without_argmax_infer_QAT_new.tar)) | Paddle Inference | CPU | 4726.65| 4134.91|None|None | 1.14 |77.37 | 71.62 |量化蒸馏训练 |

Expand Down