pool break lite(poolbreaklite臺球)

銅靈 發自 凹非寺

量子位 出品 | 公眾號 QbitAI

暑假即將到來，不用來充電學習豈不是虧大了。

有這么一份干貨，匯集了機器學習架構和模型的經典知識點，還有各種TensorFlow和PyTorch的Jupyter Notebook筆記資源，地址都在，無需等待即可取用。

除了取用方便，這份名為Deep Learning Models的資源還尤其全面。

針對每個細分知識點的介紹還尤其全面的，比如在卷積神經網絡部分，作者就由淺及深分別介紹了AlexNet、VGG、ResNet等。

干貨發布后，在GitHub短時間獲得了6000+顆星星，迅速聚集起大量人氣。

圖靈獎得主、AI大牛Yann LeCun也強烈推薦，夸贊其為一份不錯的PyTorch和TensorFlow Jupyter筆記本推薦！

這份資源的作者來頭也不小，他是威斯康星大學麥迪遜分校的助理教授Sebastian Raschka，此前還編寫過Python Machine Learning一書。

話不多說現在進入干貨時間，好東西太多篇幅較長，記得先碼后看！

原資源地址：

https://github.com/rasbt/deeplearning-models

干貨來也

1、多層感知機

多層感知機簡稱MLP，是一個打基礎的知識點：

多層感知機：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mlp/mlp-basic.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mlp/mlp-basic.ipynb

增加了Dropout部分的多層感知機：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mlp/mlp-dropout.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mlp/mlp-dropout.ipynb

具備批標準化的多層感知機：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mlp/mlp-batchnorm.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mlp/mlp-batchnorm.ipynb

從零開始了解多層感知機與反向傳播：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mlp/mlp-lowlevel.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mlp/mlp-fromscratch__sigmoid-m.ipynb

2、卷積神經網絡

在卷積神經網絡這一部分，細碎的知識點很多，包含基礎概念、全卷積網絡、AlexNet、VGG等多個內容。來看干貨：

卷積神經網絡基礎入門：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/cnn/cnn-basic.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-basic.ipynb

卷積神經網絡的初始化：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-he-init.ipynb

想用等效卷積層替代全連接的話看看下面這個：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/fc-to-conv.ipynb

全卷積神經網絡基礎知識在這里：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-allconv.ipynb

Alexnet網絡模型在CIFAR-10數據集上的實現:

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-alexnet-cifar10.ipynb

關于VGG模型，你可能需要了解VGG-16架構：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/cnn/cnn-vgg16.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-vgg16.ipynb

在CelebA上訓練的VGG-16性別分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-vgg16-celeba.ipynb

VGG19網絡架構：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-vgg19.ipynb

關于2015年被提出的經典CNN模型ResNet，最厲害的資源也在這了。

比如ResNet和殘差塊：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/resnet-ex-1.ipynb

用MNIST數據集訓練的ResNet-18數字分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet18-mnist.ipynb

用人臉屬性數據集CelebA訓練的ResNet-18性別分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet18-celeba-dataparallel.ipynb

在MNIST上訓練的ResNet-34：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet34-mnist.ipynb

在CelebA上訓練ResNet-34性別分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet34-celeba-dataparallel.ipynb

在MNIST上訓練的ResNet-50數字分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet50-mnist.ipynb

在CelebA上訓練ResNet-50性別分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet50-celeba-dataparallel.ipynb

在CelebA上訓練ResNet-101性別分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet101-celeba.ipynb

在CelebA上訓練ResNet-152性別分類器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-resnet152-celeba.ipynb

CIFAR-10分類器中的網絡：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/nin-cifar10.ipynb

3、指標學習

具有多層感知機的孿生網絡：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/metric/siame-1.ipynb

4、自編碼器

在自編碼器這一部分，同樣有很多細分類別需要學習，注意留出充足時間學習這一內容。

自編碼器的種類很多，比如全連接自編碼器：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/autoencoder/ae-basic.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-basic.ipynb

還有卷積自編碼器。比如這個反卷積（轉置卷積）卷積自編碼器：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/autoencoder/ae-deconv.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-deconv.ipynb

沒有進行池化的反卷積自編碼器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-deconv-nopool.ipynb

有最近鄰插值的卷積自編碼器：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/autoencoder/ae-conv-nneighbor.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-conv-nneighbor.ipynb

在CelebA上訓練過的有最近鄰插值的卷積自編碼器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-conv-nneighbor-celeba.ipynb

在谷歌涂鴉數據集Quickdraw上訓練過的有最近鄰插值的卷積自編碼器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-conv-nneighbor-quickdraw-1.ipynb

變分自編碼器也是自編碼器中的重要一類：

變分自編碼器基礎介紹：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-var.ipynb

卷積變分自編碼器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-conv-var.ipynb

最后，還有條件變分自編碼器也需要關注。比如在重建損失中有標簽的：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-cvae.ipynb

沒有標簽的：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-cvae_no-out-concat.ipynb

有標簽的條件變分自編碼器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-cnn-cvae.ipynb

沒有標簽的條件變分自編碼器：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/autoencoder/ae-cnn-cvae_no-out-concat.ipynb

5、生成對抗網絡（GAN）

在MNIST上的全連接GAN：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/gan/gan.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/gan/gan.ipynb

在MNIST上訓練的條件GAN：

TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/gan/gan-conv.ipynb

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/gan/gan-conv.ipynb

用Label Smoothing方法優化過的條件GAN：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/gan/gan-conv-smoothing.ipynb

6、循環神經網絡

針對多對一的情緒分析和分類問題中，包括簡單單層RNN：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/rnn_simple_imdb.ipynb

壓縮序列的簡單單層RNN：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/rnn_simple_packed_imdb.ipynb

RNN和LSTM技術：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/rnn_lstm_packed_imdb.ipynb

基于GloVe預訓練詞向量的有LSTM核的RNN：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/rnn_lstm_packed_imdb-glove.ipynb

GRU核的RNN：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/rnn_gru_packed_imdb.ipynb

多層雙向RNN：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/rnn_gru_packed_imdb.ipynb

一對多/序列到序列的生成新文本的字符RNN：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/char_rnn-charlesdickens.ipynb

7、有序回歸

針對不同場景，有三類有序回歸干貨：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/ordinal/ordinal-cnn-coral-afadlite.ipynb

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/ordinal/ordinal-cnn-niu-afadlite.ipynb

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/ordinal/ordinal-cnn-niu-afadlite.ipynb

8、方法和技巧

關于周期性學習速率，這里也有一份小技巧：

PyTorch版

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/tricks/cyclical-learning-rate.ipynb

9、PyTorch Workflow和機制

用自定義數據集加載PyTorch，這里也有一些攻略：

比如用CelebA中的人臉圖像：

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mechanics/custom-data-loader-celeba.ipynb

比如用街景數據集：

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mechanics/custom-data-loader-svhn.ipynb

比如用Quickdraw:

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mechanics/custom-data-loader-quickdraw.ipynb

在訓練和預處理環節，標準化圖像可參考：

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-standardized.ipynb

圖像信息樣本：

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/mechanics/torchvision-transform-examples.ipynb

有文本文檔的Char-RNN ：

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/rnn/char_rnn-charlesdickens.ipynb

在CelebA上訓練的VGG-16性別分類器的并行計算等：

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/cnn/cnn-vgg16-celeba-data-parallel.ipynb

10、TensorFlow Workflow與機制

這是這份干貨中的最后一個大分類，包含自定義數據集、訓練和預處理兩大部分。

內容包括：

將NumPy NPZ用于小批量訓練圖像數據集

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mechanics/image-data-chunking-npz.ipynb

用HDF5文件存儲圖像數據集，用于小規模訓練

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mechanics/image-data-chunking-hdf5.ipynb

用輸入pipeline從TFRecords文件中讀取數據

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mechanics/tfrecords.ipynb

TensorFlow數據集API

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mechanics/datat-api.ipynb

如果需要從TensorFlow Checkpoint文件和NumPy NPZ Archive中存儲和加載訓練模型，可移步：

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/mechanics/saving-and-reloading-models.ipynb

11、傳統機器學習

最后，如果你是從零開始入門，可以從傳統機器學習看起。包括感知機、邏輯回歸和Softmax回歸等。

感知機部分TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/basic-ml/perceptron.ipynb

PyTorch版筆記

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/basic-ml/perceptron.ipynb

邏輯回歸部分也是一樣：

邏輯回歸部分部分TensorFlow版Jupyter Notebooks

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/basic-ml/logistic-regression.ipynb

PyTorch版筆記

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/basic-ml/logistic-regression.ipynb

Softmax回歸，也稱為多項邏輯回歸：

Softmax回歸部分部分TensorFlow版Jupyter Notebook

https://github.com/rasbt/deeplearning-models/blob/master/tensorflow1_ipynb/basic-ml/softmax-regression.ipynb

PyTorch版筆記

https://github.com/rasbt/deeplearning-models/blob/master/pytorch_ipynb/basic-ml/softmax-regression.ipynb

傳送門

這份干貨滿滿的資源到這里就結束了，再次放上原文傳送門：

https://github.com/rasbt/deeplearning-models

超強干貨，記得收藏~

— 完 —

誠摯招聘

量子位正在招募編輯/記者，工作地點在北京中關村。期待有才氣、有熱情的同學加入我們！相關細節，請在量子位公眾號(QbitAI)對話界面，回復“招聘”兩個字。

量子位 QbitAI · 頭條號簽約作者

?'?' ? 追蹤AI技術和產品新動態