Pocketflow的簡介：https://pocketflow.github.io/

由兩個modules組成：Learner module 負責利用隨機hyperparameters壓縮并fine-tuning，optimizer module 負責依據隨機參數對應的結果確定最佳的壓縮參數。Learn module 遍歷的hyperparameter來自于：

NameDescription

ChannelPrunedLearner	channel pruning with LASSO-based channel selection (He et al., 2017)
DisChnPrunedLearner	discrimination-aware channel pruning (Zhuang et al., 2018)
WeightSparseLearner	weight sparsification with dynamic pruning schedule (Zhu & Gupta, 2017)
UniformQuantLearner	weight quantization with uniform reconstruction levels (Jacob et al., 2018)
UniformQuantTFLearner	weight quantization with uniform reconstruction levels and TensorFlow APIs
NonUniformQuantLearner	weight quantization with non-uniform reconstruction levels (Han et al., 2016)

壓縮后可以快速訓練收斂。

optimizer module 的策略來自于

Gaussian Processes (GP, Mockus, 1975), Tree-structured Parzen Estimator (TPE, Bergstra et al., 2013), and Deterministic Deep Policy Gradients (DDPG, Lillicrap et al., 2016) 以及當前hyperparameters的reward

安裝Pocketflow：https://pocketflow.github.io/installation/

本地安裝或在docker環境中安裝

本地安裝：

配置文件 path.conf 說明數據集路徑(.tfrecord)以及預訓練模型的路徑

首先安裝 anaconda 和 tensorflow，注意python 版本不能高于3.6

一句話就可訓練：./scripts/run_local.sh nets/resnet_at_cifar10_run.py

docker環境安裝：

官方推薦??horovod?docker ，因其支持多GPU分布

一句話訓練：./scripts/run_docker.sh nets/resnet_at_cifar10_run.py

?

壓縮方法：

參考https://pocketflow.github.io/tutorial/

輸出：*.tflite 文件

Step 1: 準備數據成TFRecord格式

Step 2: 根據上面的步驟訓練或直接下載 .ckpt 模型 保存在?models

Step 3: 選擇壓縮方法 開始調優壓縮后的模型?

./scripts/run_local.sh nets/resnet_at_ilsvrc12_run.py \ --learner dis-chn-pruned

壓縮方法與arg對應：

Learner nameLearner classNote

full-prec	FullPrecLearner	No model compression
channel	ChannelPrunedLearner	Channel pruning with LASSO-based channel selection (He et al., 2017)
dis-chn-pruned	DisChnPrunedLearner	Discrimination-aware channel pruning (Zhuang et al., 2018)
weight-sparse	WeightSparseLearner	Weight sparsification with dynamic pruning schedule (Zhu & Gupta, 2017)
uniform	UniformQuantLearner	Weight quantization with uniform reconstruction levels (Jacob et al., 2018)
uniform-tf	UniformQuantTFLearner	Weight quantization with uniform reconstruction levels and TensorFlow APIs
non-uniform	NonUniformQuantLearner	Weight quantization with non-uniform reconstruction levels (Han et al., 2016)

保存在models_dcp_eval文件夾

Step 3: 轉化成 tflite 格式：

python tools/conversion/export_pb_tflite_models.py \ --model_dir models_dcp_eval

保存在?models_dcp_eval

?

?

?

總結

以上是生活随笔為你收集整理的Pocketflow 使用的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。