# TensorFlowMnist

**Repository Path**: jacen789/TensorFlowMnist

## Basic Information

- **Project Name**: TensorFlowMnist
- **Description**: 利用神经网络识别手写数字（MNIST 数据集）
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2019-12-16
- **Last Updated**: 2020-12-19

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# TensorFlowMnist
利用神经网络识别手写数字（MNIST 数据集）

## 来源:
https://github.com/GoogleCloudPlatform/tensorflow-without-a-phd/tree/master/tensorflow-mnist-tutorial

## 让模型循环训练1001次后模型的accuracy和loss
**mnist_1.0_softmax**
> test accuracy:0.9174 test loss: 29.2893

**mnist_2.0_five_layers_sigmoid**
> test accuracy:0.9569 test loss: 16.2656

**mnist_2.1_five_layers_relu_lrdecay**
> test accuracy:0.9669 test loss: 10.4956

**mnist_2.2_five_layers_relu_lrdecay_dropout**
> test accuracy:0.9653 test loss: 12.5906

**mnist_3.0_convolutional**
> test accuracy:0.9838 test loss: 5.20661

**mnist_3.1_convolutional_bigger_dropout**
> test accuracy0.9883 test loss 3.61283

**mnist_4.0_batchnorm_five_layers_sigmoid**
> test accuracy:0.9547 test loss: 15.8005

**mnist_4.1_batchnorm_five_layers_relu**
> test accuracy:0.9771 test loss: 8.02166

**mnist_4.2_batchnorm_convolutional**
> test accuracy:0.9876 test loss: 4.3569