# NeurstrucEnergy **Repository Path**: neucloudlab/NeurstrucEnergy ## Basic Information - **Project Name**: NeurstrucEnergy - **Description**: NeurstrucEnergy - **Primary Language**: Python - **License**: MIT - **Default Branch**: main - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2022-10-14 - **Last Updated**: 2022-10-14 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # NeurstrucEnergy |-- NeurstrucEnergy   |-- .DS_Store   |-- .gitignore   |-- LICENSE   |-- README.md   |-- utlis.py   |-- CNNGeneration   |  |-- CustomCNNGeneration.py   |  |-- RandomCNNGeneration.py   |-- data   |  |-- .DS_Store   |  |-- conv2d_energy_profiling.py   |  |-- energy_profiling.py   |  |-- Monitor.py   |  |-- dataset   |    |-- .DS_Store   |    |-- conv2d.txt   |    |-- conv2d_PreVIousNet.txt   |    |-- custom_testset_1.txt   |    |-- custom_testset_2.txt   |    |-- custom_testset_3.txt   |    |-- NNtest.py   |    |-- random_testset_1.txt   |    |-- random_testset_2.txt   |    |-- random_testset_3.txt   |-- model   |  |-- .DS_Store   |  |-- BiGNN.py   |  |-- train.py   |  |-- Transformer.py   |  |-- PerLayerRegression   |    |-- NeuralPower_PolyModel.py   |    |-- PreVIous_LinearModel.py   |-- TrainedModel     |-- .DS_Store     |-- CustomPredict.py     |-- RandomPredict.py     |-- CustomModel     |  |-- .DS_Store     |  |-- model_592     |-- RandomModel        |-- model_432 ### File Description 文件说明 File | Description ---|--- utlis.py | Some functions utilized by other codes. (共用函数) CustomCNNGeneration.py | Customized generation algorithm (订制生成算法) RandomCNNGeneration.py | Random generation algorithm(随机生成算法) conv2d\_energy\_profiling.py | Obtain inference energy consumption of conv2d layer. (得到conv2d层的推理能耗) energy\_profiling.py | Obtain inference energy consumption of networks. (得到神经网络的推理能耗) Monitor.py | Connect with the power meter. (与功率计通信) conv2d.txt | 1000 convolution layers from the randomly generated dataset. (从随机生成数据集中抽取的1000个卷积层) conv2d_PreVIousNet.txt | Conv2d layers dataset utilized by PreVIous. (PreVIous的Conv2d层数据集) custom\_testset\_*.txt | The 5 test nets for the experiment on custom dataset. (定制数据集实验中的5个测试网络) random\_testset\_*.txt | The 17 test nets for the experiment on random dataset. (随机数据集实验中的17个测试网络) BiGNN.py | An implemention of bi-directional GNN. (双向GNN实现) train.py | Train our bi-directional GNN model. (训练模型) Transformer.py | An implemention of Transformer. (Trainsformer实现) NeuralPower_PolyModel.py | The experiment on polynomial regression model for layer regression. (多项式模型预测层能耗实验) PreVIous_LinearModel.py | The experiment on linear regression model for layer regression. (线性回归模型预测层能耗实验) CustomPredict.py | The experiment on custom dataset. (定制数据集实验) RandomPredict.py | The experiment on random dataset. (随机数据集实验) model_592 | The parameters of model trained on custom dataset. (定制数据集模型参数) model_432 | The parameters of model trained on random dataset. (随机数据集模型参数) ### Model Training 训练模型 ##### Trained on randomly generated dataset ``` python train.py ``` ##### Trained on custom dataset ``` python train.py --hidden 80 --reverse-hidden 16 --lr 0.00045 --dataset custom ``` ### Reproduce the experimental results 复现实验结果 We only provide the parameters of bi-directional model for one epoch in this depository, and the download addresses for all 50 consecutive epochs and the complete dataset are as follows: 在本仓库中我们只提供训练过程中1轮的模型参数,全部连续50轮的模型参数和完整数据集的下载地址如下: > https://drive.google.com/drive/folders/1THfcWXFy_EFYByZPDoRLth7sMspQaLEM?usp=sharing > 链接: https://pan.baidu.com/s/1FTFTu9ocspmFkU6_1QabtA 提取码: jfli ##### Trained on randomly generated dataset ``` python RandomPredict.py ``` ##### Trained on custom dataset ``` python CustomPredict.py ```