Optimize doc, test=document_fix ()

pull/646/head
Tingquan Gao 2021-03-16 16:12:34 +08:00 committed by GitHub
parent 7f3f44c105
commit 9a0d8af0ab
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 7 additions and 7 deletions
docs
en/tutorials
zh_CN/tutorials

View File

@ -171,14 +171,15 @@ python3 tools/train.py -c ./configs/quick_start/ResNet50_vd_ssld_random_erasin
It improves by 1.27\% to 96.27\%
* Save ResNet50_vd pretrained model to experience next chapter.
### Distillation
* The ResNet50_vd model pretrained on previous chapter will be used as the teacher model to train student model. Save the model to specified directory, command as follows:
```shell
cp -r output/ResNet50_vd/19/ ./pretrained/flowers102_R50_vd_final/
```
### Distillation
* Use `extra_list.txt` as unlabeled data, Note:
* Samples in the `extra_list.txt` and `val_list.txt` don't have intersection
* Because of in the source code, label information is unused, This is still unlabeled distillation

View File

@ -166,14 +166,13 @@ python3 tools/train.py -c ./configs/quick_start/ResNet50_vd_ssld_random_erasi
最终flowers102验证集上的精度为0.9627使用数据增广可以使得模型精度再次提升1.27\%。
* 如果希望体验`3.6节`的知识蒸馏部分可以首先保存训练得到的ResNet50_vd预训练模型到合适的位置作为蒸馏时教师模型的预训练模型。脚本如下所示。
### 3.6 知识蒸馏小试牛刀
* 本小节将尝试使用知识蒸馏技术对MobileNetV3_large_x1_0模型进行训练使用`3.5小节`训练得到的ResNet50_vd模型作为蒸馏所用的教师模型首先将`3.5小节`训练得到的ResNet50_vd模型保存到指定目录脚本如下。
```shell
cp -r output/ResNet50_vd/best_model/ ./pretrained/flowers102_R50_vd_final/
```
### 3.6 知识蒸馏小试牛刀
* 使用flowers102数据集进行模型蒸馏为了进一步提提升模型的精度使用`extra_list.txt`充当无标签数据,在这里有几点需要注意:
* `extra_list.txt`与`val_list.txt`的样本没有重复,因此可以用于扩充知识蒸馏任务的训练数据。
* 即使引入了有标签的extra_list.txt中的图像但是代码中没有使用标签信息因此仍然可以视为无标签的模型蒸馏。