TensorLayer 2.0

[zsdonghao]

NETWORK API REFACTORING - TO DO LIST

Design Docs

Refactoring Codes

Dear Contributors,

@DEKHTIARJonathan @akaraspt @luomai @lgarithm @JingqingZ @fangde etal.

As we discussed previously, TensorLayer 2.0 should support both eager and graph mode. The new API design is here https://github.com/luomai/tensorlayer2-design/issues/7

To make the refactoring faster, I simply fork tensorlayer/tensorlayer into zsdonghao/tensorlayer2: https://github.com/zsdonghao/tensorlayer2 , we can merge the branch back to tensorlayer/tensorlayer when the refactoring is finished. In doing so, the contributions will be in may commits rather than only 1.

Work to be done

Layers

• [x] core.py:
  • Layer:
  • [x] refactored @JingqingZ 2019/01/28
  • [x] tested @JingqingZ 2019/01/31 2019/03/06
  • [x] documentation @JingqingZ 2019/03/06
  • ModelLayer:
  • [x] created @JingqingZ 2019/01/28
  • [x] tested @JingqingZ 2019/03/06
  • [x] documentation @JingqingZ 2019/03/06
  • LayerList:
  • [x] created @JingqingZ 2019/01/28 @ChrisWu1997
  • [x] tested @JingqingZ 2019/03/06
  • [x] documentation @JingqingZ 2019/03/06
  • LayerNode:
  • [x] created @ChrisWu1997
  • [x] tested @ChrisWu1997 2019/03/22
  • [x] documentation @ChrisWu1997 2019/03/22
• [x] activation.py:
  • PRelu:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/20
  • [x] tested @JingqingZ 2019/03/20
  • [x] documentation @JingqingZ 2019/03/20
  • PRelu6:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/20
  • [x] tested @JingqingZ 2019/03/20
  • [x] documentation @JingqingZ 2019/03/20
  • PTRelu6:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/20
  • [x] tested @JingqingZ 2019/03/20
  • [x] documentation @JingqingZ 2019/03/20
• convolution/
  • AtrousConv1dLayer, AtrousConv2dLayer and AtrousDeConv2d are removed, use Conv1d/2d and DeConv2d with dilation_rate instead. (🀄️remember to change CN docs)
  • BinaryConv2d:
  • [x] refactored @zsdonghao 2018/12/05
  • [x] tested @warshallrho 2019/03/16
  • [x] documentation @warshallrho 2019/03/20
  • Conv1d:
  • [x] refactored @zsdonghao 2019/01/16
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • Conv2d:
  • [x] refactored @zsdonghao 2019/01/16
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • Conv3d:
  • [x] add @zsdonghao 2019/01/16 : (🀄️remember to change CN docs)
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • Conv1dLayer:
  • [x] refactored @zsdonghao 2018/12/05
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • Conv2dLayer:
  • [x] refactored @zsdonghao 2018/12/05
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • Conv3dLayer:
  • [x] refactored @zsdonghao 2018/12/05
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • DeConv1dLayer:
  • [x] refactored @warshallrho 2019/03/16
  • [x] tested @warshallrho 2019/03/16
  • [x] documentation @warshallrho 2019/03/17
  • DeConv2dLayer:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • DeConv3dLayer:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • DeConv2d:
  • [x] refactored @zsdonghao 2019/01/16
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • DeConv3d:
  • [x] refactored @zsdonghao 2019/01/16
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/17
  • DeformableConv2d:
  • [x] refactored @warshallrho 2019/03/18
  • [x] tested @warshallrho 2019/03/18
  • [x] documentation @warshallrho 2019/03/18
  • DepthwiseConv2d:
  • [x] refactored @zsdonghao 2018/12/05
  • [x] tested @warshallrho 2019/03/15
  • [x] documentation @warshallrho 2019/03/18
  • DorefaConv2d:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @warshallrho 2019/03/17
  • [x] documentation @warshallrho 2019/03/20
  • GroupConv2d:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @warshallrho 2019/03/17
  • [x] documentation @warshallrho 2019/03/20
  • QuanConv2d:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @warshallrho 2019/03/17
  • [x] documentation @warshallrho 2019/03/20
  • QuanConv2dWithBN:
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
  • SeparableConv1d:
  • [x] refactored @zsdonghao 2019/01/16
  • [x] tested @warshallrho 2019/03/17
  • [x] documentation @warshallrho 2019/03/18
  • SeparableConv2d:
  • [x] refactored @zsdonghao 2019/01/16
  • [x] tested @warshallrho 2019/03/17
  • [x] documentation @warshallrho 2019/03/18
  • SubpixelConv1d:
  • [x] refactored @zsdonghao 2018/12/05 @warshallrho 2019/03/18
  • [x] tested @warshallrho 2019/03/18
  • [x] documentation @warshallrho 2019/03/18
  • SubpixelConv2d:
  • [x] refactored @zsdonghao 2018/12/05 @warshallrho 2019/03/18
  • [x] tested @warshallrho 2019/03/18
  • [x] documentation @warshallrho 2019/03/18
  • TernaryConv2d:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @warshallrho 2019/03/17
  • [x] documentation @warshallrho 2019/03/20
• dense/ [WIP] @ChrisWu1997
  • BinaryDense:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @ChrisWu1997 2019/04/23 need further test by example
  • [x] documentation @ChrisWu1997 2019/04/23
  • Dense:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/01/28
  • [x] tested @JingqingZ 2019/01/31 2019/03/06 2019/03/15
  • [x] documentation @JingqingZ 2019/03/15
  • DorefaDense:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @ChrisWu1997 2019/04/23 need further test by example
  • [x] documentation @ChrisWu1997 2019/04/23
  • DropconnectDense:
  • [x] refactored @zsdonghao 2018/12/05
  • [x] tested @ChrisWu1997 2019/04/23 need further test by example
  • [x] documentation @ChrisWu1997 2019/04/23
  • QuanDense:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @ChrisWu1997 2019/04/23 need further test by example
  • [x] documentation @ChrisWu1997 2019/04/23
  • QuanDenseWithBN:
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
  • TernaryDense:
  • [x] refactored @zsdonghao 2018/12/06
  • [x] tested @ChrisWu1997 2019/04/23 need further test by example
  • [x] documentation @ChrisWu1997 2019/04/23
• dropout.py
  • Dropout:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/01/28
  • [x] tested @JingqingZ 2019/01/31 2019/03/06 2019/03/15
  • [x] documentation @JingqingZ 2019/03/15
• extend.py
  • ExpandDims:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/22
  • [x] tested @JingqingZ 2019/03/22
  • [x] documentation @JingqingZ 2019/03/22
  • Tile:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/22
  • [x] tested @JingqingZ 2019/03/22
  • [x] documentation @JingqingZ 2019/03/22
• image_resampling.py
  • UpSampling2d:
  • [x] refactored @zsdonghao 2018/12/04 @ChrisWu1997 2019/04/03
  • [x] tested @ChrisWu1997 2019/04/03
  • [x] documentation @ChrisWu1997 2019/04/03
  • DownSampling2d:
  • [x] refactored @zsdonghao 2018/12/04 @ChrisWu1997 2019/04/03
  • [x] tested @ChrisWu1997 2019/04/03
  • [x] documentation @ChrisWu1997 2019/04/03
• importer.py
  • SlimNets:
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
  • Keras:
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
• inputs.py
  • Input:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/01/28
  • [x] tested @JingqingZ 2019/03/06
  • [x] documentation @JingqingZ 2019/03/06
• embedding.py
  • OneHotInput: --> OneHot (🀄️remember to change CN docs)
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/02/23
  • [x] tested @JingqingZ 2019/03/19
  • [x] documentation @JingqingZ 2019/03/19
  • Word2vecEmbeddingInput: --> Word2vecEmbedding (🀄️remember to change CN docs)
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/02/21
  • [x] tested @JingqingZ 2019/03/19
  • [x] documentation @JingqingZ 2019/03/19
  • EmbeddingInput: --> Embedding
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/02/22
  • [x] tested @JingqingZ 2019/03/19
  • [x] documentation @JingqingZ 2019/03/19
  • AverageEmbeddingInput: --> AverageEmbedding (🀄️remember to change CN docs)
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/02/20
  • [x] tested @JingqingZ 2019/03/19
  • [x] documentation @JingqingZ 2019/03/19
• lambda_layers.py
  • ElementwiseLambda:
  • [x] refactored @JingqingZ 2019/03/24
  • [x] tested @JingqingZ 2019/03/24
  • [x] documentation @JingqingZ 2019/03/24
  • Lambda:
  • [x] refactored @JingqingZ 2019/03/24
  • [x] tested @JingqingZ 2019/03/24
  • [x] documentation @JingqingZ 2019/03/24
• merge.py
  • Concat:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @JingqingZ 2019/03/15
  • [x] documentation @JingqingZ 2019/03/15
  • Elementwise:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/15
  • [x] tested @JingqingZ 2019/03/15
  • [x] documentation @JingqingZ 2019/03/15
• noise.py
  • GaussianNoise:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @warshallrho 2019/03/20
  • [x] documentation @warshallrho 2019/03/20
• normalization.py
  • BatchNorm:
  • [x] refactored @ChrisWu1997 2019/01/22 @ChrisWu1997 2019/03/05
  • [x] tested @ChrisWu1997 2019/03/22
  • [x] documentation @ChrisWu1997 2019/03/22
  • BatchNorm1d:
  • [x] refactored @ChrisWu1997 2019/03/05
  • [x] tested @ChrisWu1997 2019/03/22
  • [x] documentation @ChrisWu1997 2019/03/22
  • BatchNorm2d:
  • [x] refactored @ChrisWu1997 2019/03/05
  • [x] tested @ChrisWu1997 2019/03/22
  • [x] documentation @ChrisWu1997 2019/03/22
  • BatchNorm3d:
  • [x] refactored @ChrisWu1997 2019/03/05
  • [x] tested @ChrisWu1997 2019/03/22
  • [x] documentation @ChrisWu1997 2019/03/22
  • GroupNorm:
  • [x] refactored @zsdonghao 2018/12/05
  • [ ] tested
  • [ ] documentation
  • InstanceNorm:
  • [x] refactored @zsdonghao 2018/12/05
  • [ ] tested
  • [ ] documentation
  • LayerNorm:
  • [x] refactored @ChrisWu1997 2019/01/23
  • [ ] tested
  • [ ] documentation
  • LocalResponseNorm:
  • [x] refactored @zsdonghao 2018/12/05
  • [ ] tested
  • [ ] documentation
  • SwitchNorm:
  • [x] refactored @zsdonghao 2018/12/05
  • [ ] tested
  • [ ] documentation
• padding.py
  • PadLayer:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @warshallrho 2019/03/21
  • [x] documentation @warshallrho 2019/03/21
  • ZeroPad1d:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @warshallrho 2019/03/21
  • [x] documentation @warshallrho 2019/03/21
  • ZeroPad2d:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @warshallrho 2019/03/21
  • [x] documentation @warshallrho 2019/03/21
  • ZeroPad3d:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @warshallrho 2019/03/21
  • [x] documentation @warshallrho 2019/03/21
• pooling/
  • MaxPool1d:
    • [x] refactored @zsdonghao 2019/01/08
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/19
  • MaxPool2d:
    • [x] refactored @zsdonghao 2018/12/06
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/19
  • MaxPool3d:
    • [x] refactored @zsdonghao 2019/01/08
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/19
  • MeanPool1d:
    • [x] refactored @zsdonghao 2019/01/08
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/19
  • MeanPool2d:
    • [x] refactored @zsdonghao 2019/01/08
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/19
  • MeanPool3d:
    • [x] refactored @zsdonghao 2019/01/08
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/19
  • GlobalMaxPool1d:
    • [x] refactored @zsdonghao 2018/12/06
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/15
  • GlobalMaxPool2d:
    • [x] refactored @zsdonghao 2018/12/06
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/15
  • GlobalMaxPool3d:
    • [x] refactored @zsdonghao 2018/12/06
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/15
  • GlobalMeanPool1d:
    • [x] refactored @zsdonghao 2018/12/06
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/15
  • GlobalMeanPool2d:
    • [x] refactored @zsdonghao 2018/12/06
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/15
  • GlobalMeanPool3d:
    • [x] refactored @zsdonghao 2018/12/06
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/15
  • PoolLayer:
    • [x] refactored @zsdonghao 2018/12/04
    • [x] tested @warshallrho 2019/03/15
    • [x] documentation @warshallrho 2019/03/18
• quantize_layers.py
  • Sign:
  • [x] refactored
  • [ ] tested
  • [ ] documentation
• recurrent/
  • BiRNN:
  • [x] refactored @JingqingZ 2019/04/08
  • [x] tested @JingqingZ 2019/04/08
  • [x] documentation @JingqingZ 2019/04/08
  • ConvLSTM:
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
  • RNN:
  • [x] refactored @JingqingZ 2019/03/31
  • [x] tested @JingqingZ 2019/03/31
  • [x] documentation @JingqingZ 2019/03/31
  • Seq2Seq:
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
• shape.py
  • Flatten:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/22
  • [x] tested @JingqingZ 2019/03/22
  • [x] documentation @JingqingZ 2019/03/22
  • Reshape:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/22
  • [x] tested @JingqingZ 2019/03/22
  • [x] documentation @JingqingZ 2019/03/22
  • Transpose:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/22
  • [x] tested @JingqingZ 2019/03/22
  • [x] documentation @JingqingZ 2019/03/22
• scale.py
  • Scale:
  • [x] refactored @zsdonghao 2018/12/04 @JingqingZ 2019/03/22
  • [x] tested @JingqingZ 2019/03/22
  • [x] documentation @JingqingZ 2019/03/22
• contrib
  • ROIPooling:
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
• spatial_transformer.py
  • SpatialTransformer2dAffine: see test_layers_spatial_transformer.py
  • [ ] refactored
  • [ ] tested
  • [ ] documentation
• stack.py [WIP] @ChrisWu1997
  • Stack:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @ChrisWu1997 2019/04/23
  • [x] documentation @ChrisWu1997 2019/04/23
  • UnStack:
  • [x] refactored @zsdonghao 2018/12/04
  • [x] tested @ChrisWu1997 2019/04/23
  • [x] documentation @ChrisWu1997 2019/04/23
• time_distribution.py Remove, as eager mode support this feature (🀄️remember to change CN docs)
  • TimeDistributed:

tl.models

• core.py
  • Model:
  • [x] refactored @JingqingZ 2019/01/28 @ChrisWu1997 2019/02/16 2019/02/22
  • [x] tested @ChrisWu1997 2019/03/21
  • [x] documentation @ChrisWu1997 2019/03/21
• vgg.py
  • vgg:
  • [x] refactored @warshallrho 2019/02/19
  • [ ] tested
  • [x] documentation @warshallrho 2019/03/21 @ChrisWu1997 2019/03/21
  • vgg16:
  • [x] refactored @warshallrho 2019/02/19
  • [ ] tested
  • [x] documentation @warshallrho 2019/03/21 @ChrisWu1997 2019/03/21
  • vgg19:
  • [x] refactored @warshallrho 2019/03/09
  • [ ] tested
  • [x] documentation @warshallrho 2019/03/21 @ChrisWu1997 2019/03/21
• mobilenetv1.py
  • MobileNet:
  • [x] refactored @ChrisWu1997 2019/04/23
  • [x] tested @ChrisWu1997 2019/04/23
  • [x] documentation @ChrisWu1997 2019/04/23
  • SqueezeNet:
  • [x] refactored @ChrisWu1997 2019/04/23
  • [x] tested @ChrisWu1997 2019/04/23
  • [x] documentation @ChrisWu1997 2019/04/23

Examples

• basic_tutorials Too many basic tutorials, some codes can be removed.
  • [x] Static model example MNIST @JingqingZ 2019/01/28 2019/03/24
  • [x] Dynamic model example MNIST @JingqingZ 2019/01/28 2019/03/24
  • [x] Static model example CIFAR10 (with dataset API) @ChrisWu1997 2019/03/24
  • [x] Siamese example MNIST @ChrisWu1997 2019/03/26
  • tutorial_mnist_float16.py removed by @ChrisWu1997
  • tutorial_mnist_simple.py removed by @ChrisWu1997
• data_process
  • tutorial_fast_affine_transform.py
    • [x] refactored @ChrisWu1997 2019/04/11
    • [x] tested @ChrisWu1997 2019/04/11
  • tutorial_image_preprocess.py removed by @zsdonghao
  • tutorial_tf_dataset_voc.py
    • [x] refactored @ChrisWu1997 2019/04/11
    • [x] tested @ChrisWu1997 2019/04/11
  • tutorial_tfrecord.py
    • [x] refactored @ChrisWu1997 2019/04/11
    • [x] tested @ChrisWu1997 2019/04/11
  • tutorial_tfrecord2.py
    • [x] refactored @ChrisWu1997 2019/04/11
    • [x] tested @ChrisWu1997 2019/04/11
  • tutorial_tfrecord3.py
    • [ ] refactored
    • [ ] tested
• database
  • [ ] refactored
  • [ ] tested
• distributed_training
  • tutorial_cifar10_distributed_trainer.py
    • [ ] refactored
    • [ ] tested
  • tutorial_mnist_distributed_trainer.py
    • [ ] refactored
    • [ ] tested
• keras_tfslim
  • tutorial_keras.py
    • [x] refactored @ChrisWu1997 2019/04/11
    • [x] tested @ChrisWu1997 2019/04/11
  • tutorial_tfslim.py removed by @ChrisWu1997
• pretrained_cnn
  • tutorial_inceptionV3_tfslim.py
  • tutorial_mobilenet.py removed by @ChrisWu1997 2019/04/23
  • tutorial_models_mobilenetv1.py
    • [x] refactored @ChrisWu1997 2019/04/23
    • [x] tested @ChrisWu1997 2019/04/23
  • tutorial_models_squeezenetv1.py
    • [x] refactored @ChrisWu1997 2019/04/23
    • [x] tested @ChrisWu1997 2019/04/23
  • tutorial_models_vgg.py
    • [x] refactored @warshallrho 2019/04/30
    • [ ] tested
  • tutorial_models_vgg_static.py
    • [x] refactored @warshallrho 2019/04/30
    • [ ] tested
  • tutorial_models_vgg16.py
    • [x] refactored @warshallrho 2019/02/19
    • [ ] tested
  • tutorial_models_vgg19.py
    • [x] refactored @warshallrho 2019/03/09
    • [ ] tested
  • tutorial_squeezenet.py removed by @ChrisWu1997 2019/04/23
  • tutorial_vgg16.py removed by @warshallrho 2019/04/30
  • tutorial_vgg19.py removed by @warshallrho 2019/04/30
• quantized_net
  • tutorial_binarynet_cifar10_tfrecord.py
    • [x] refactored
    • [x] tested
  • tutorial_binarynet_mnist_cnn.py
    • [x] refactored
    • [x] tested
  • tutorial_dorefanet_cifar10_tfrecord.py
    • [x] refactored
    • [x] tested
  • tutorial_dorefanet_mnist_cnn.py
    • [x] refactored
    • [x] tested
  • tutorial_quanconv_cifar10.py
    • [x] refactored
    • [x] tested
  • tutorial_quanconv_mnist.py
    • [x] refactored
    • [x] tested
  • tutorial_ternaryweight_cifar10_tfrecord.py
    • [x] refactored
    • [x] tested
  • tutorial_ternaryweight_mnist_cnn.py
    • [x] refactored
    • [x] tested
• reinforcement_learning
  • tutorial_atari_pong.py @zsdonghao 2019/01/21
    • [x] refactored
    • [x] tested
  • tutorial_bipedalwalker_a3c_continuous_action.py
    • [ ] refactored
    • [ ] tested
  • tutorial_cartpole_ac.py @zsdonghao 2019/02/17
    • [x] refactored
    • [x] tested
  • tutorial_frozenlake_dqn.py @zsdonghao 2019/02/16
    • [x] refactored
    • [x] tested
  • tutorial_frozenlake_q_table.py @zsdonghao 2019/02/16
    • [x] refactored
    • [x] tested
• text_classification
  • tutorial_imdb_fasttext.py @JingqingZ 2019/03/14
    • [x] refactored
    • [x] tested
• text_generation
  • tutorial_generate_text.py
    • [ ] refactored
    • [ ] tested
• text_ptb Are they duplicated?
  • tutorial_ptb_lstm_state_is_tuple.py
    • [ ] refactored
    • [ ] tested
  • tutorial_ptb_lstm.py
    • [ ] refactored
    • [ ] tested
• text_word_embedding
  • tutorial_word2vec_basic.py @JingqingZ 2019/02/21 2019/03/19
    • [x] refactored
    • [x] tested

Others

• tl.activation.py
  • [x] refactored @JingqingZ 2019/03/06
  • [x] tested @JingqingZ 2019/03/06
  • [x] documentation @JingqingZ 2019/03/06
• tl.cli
  • [x] refactored no update needed @ChrisWu1997 2019/04/12
• tl.decorators
  • [x] refactored no update needed @ChrisWu1997 2019/04/12
• tl.logging
  • [x] refactored no update needed @ChrisWu1997 2019/04/12
• tl.optimizers
  • [ ] refactored
• tl.third_party
  • [ ] refactored
• tl.array_ops
  • [x] refactored no update needed @ChrisWu1997 2019/04/12
• tl.cost
  • [x] refactored @ChrisWu1997 2019/04/12
  • [x] documentation @ChrisWu1997 2019/04/12
• tl.db [WIP] @ChrisWu1997
  • [ ] refactored
• tl.distributed
  • [ ] refactored
• tl.initializers
  • [x] refactored @ChrisWu1997 2019/04/12
  • [x] tested @ChrisWu1997 2019/04/12
  • [x] documentation @ChrisWu1997 2019/04/12
• tl.iterate
  • [x] refactored no update needed @ChrisWu1997 2019/04/12
• tl.lazy_imports
  • [x] refactored no update needed @ChrisWu1997 2019/04/12
• tl.nlp @OliverZijia @JingqingZ
  • [x] refactored
• tl.package_info
  • [ ] refactored
• tl.prepro
  • [x] refactored @ChrisWu1997 2019/04/11
• tl.rein
  • [ ] refactored
• tl.utils
  • [x] refactored @ChrisWu1997 2019/04/17
  • [x] tested _by tutorial_mnist_simple.py_ @ChrisWu1997 2019/04/17
  • [x] documentation @ChrisWu1997 2019/04/17
• tl.visualize
  • [x] refactored no update needed @ChrisWu1997 2019/04/12

Unittests Status:

• performance_test
  • VGG @JingqingZ @ChrisWu1997 @warshallrho 2019/03/20
• layers
  • test_layernode.py @ChrisWu1997 2019/03/22
  • test_layers_activation.py @JingqingZ 2019/03/20
  • test_layers_convolution.py (1d, 2d, 3d) @warshallrho 2019/03/20
  • test_layers_core_basedense_dropout.py @JingqingZ 2019/03/06
  • test_layers_convolution_deformable.py @warshallrho 2019/03/18
  • test_layers_embedding.py @JingqingZ 2019/03/19
  • test_layers_extend.py @JingqingZ 2019/03/22
  • test_layers_lambda.py @JingqingZ 2019/03/24
  • test_layers_merge.py @JingqingZ 2019/03/15
  • test_layers_noise.py @warshallrho 2019/03/21
  • test_layers_padding.py @warshallrho 2019/03/21
  • test_layers_pooling.py @warshallrho 2019/03/18
  • test_layers_recurrent.py @JingqingZ 2019/03/06
  • test_layers_scale.py @JingqingZ 2019/03/22
  • test_layers_shape.py @JingqingZ 2019/03/22
• test_activations.py @JingqingZ 2019/03/06
• models
  • test_model_save_graph.py @warshallrho 2019/04/30

Unittests Status (Pending):

Some testing codes can be removed.

• test_array_ops.py
• test_decorators.py
• test_documentation.py
• test_layers_basic.py
• test_layers_flow_control.py removed in favour of eager mode @zsdonghao 2018/12/04 (🀄️remember to change CN docs)
• test_layers_importer.py
• test_layers_normalization.py
• test_layers_padding.py
• test_layers_spatial_transformer.py
• test_layers_stack.py
• test_layers_super_resolution.py
• test_layers_time_distributed.py
• test_logging.py
• test_logging_hyperdash.py
• test_mnist_simple.py
• test_model_compilednetwork.py
• test_models.py
• test_network_custom_2d.py
• test_network_custom_input_layers.py
• test_network_custom_multiple_inputs.py
• test_network_custom_multiple_outputs.py
• test_network_sequential_1d.py
• test_network_sequential_2d.py
• test_network_sequential_3d.py
• test_network_sequential_rnn.py
• test_optimizer_amsgrad.py
• test_pydocstyle.py
• test_reuse_mlp.py
• test_tf_layers.py
• test_timeout.py
• test_utils_predict.py
• test_yapf_format.py

tl.files

All save/load methods are also wrapped as class method in model core.

• save_hdf5_graph
  • [x] created @warshallrho 2019/04/27
  • [x] tested @warshallrho 2019/04/27
  • [x] documentation @warshallrho 2019/04/27
• load_hdf5_graph
  • [x] created @warshallrho 2019/04/27
  • [x] tested @warshallrho 2019/04/27
  • [x] documentation @warshallrho 2019/04/27
• save_weights_to_hdf5
  • [x] created
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• load_hdf5_to_weights_in_order
  • [x] created
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• load_hdf5_to_weights
  • [x] created
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• save_npz([save_list, name, sess]) @ChrisWu1997 2019/02/21 --> save_npz([save_list, name]) @ChrisWu1997 2019/03/21
  • [x] refactored
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• load_npz([path, name]) @ChrisWu1997 2019/02/21
  • [x] refactored
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• assign_params(sess, params, network) --> assign_weights (🀄️remember to change CN docs) @ChrisWu1997 2019/02/22
  • [x] refactored
  • [ ] tested
• load_and_assign_npz([sess, name, network]) @ChrisWu1997 2019/02/21 --> load_and_assign_npz([name, network]) @ChrisWu1997 2019/03/21
  • [x] refactored
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• save_npz_dict([save_list, name, sess]) @ChrisWu1997 2019/02/22 --> save_npz_dict([save_list, name]) @ChrisWu1997 2019/03/21
  • [x] refactored
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• load_and_assign_npz_dict([name, sess]) --> ([name, network]) @ChrisWu1997 2019/03/21
  • [x] refactored
  • [x] tested @ChrisWu1997 2019/03/26
  • [x] documentation @ChrisWu1997 2019/03/26
• save_ckpt([sess, mode_name, save_dir, …]) @ChrisWu1997 2019/02/22
  • [x] refactored
  • [ ] tested
• load_ckpt([sess, mode_name, save_dir, …]) @ChrisWu1997 2019/02/22
  • [x] refactored
  • [ ] tested

[zsdonghao]

after running tf_upgrade_v2 for layers, some initialisations need manual check:

Detected 109 errors that require attention
--------------------------------------------------------------------------------
tensorlayer/layers/recurrent.py:151:24: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/recurrent.py:336:24: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/recurrent.py:640:18: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/recurrent.py:642:18: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/recurrent.py:646:61: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/recurrent.py:715:24: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/recurrent.py:913:50: tf.zeros_like requires manual check: WARNING: tf.zeros_like and tf.ones_like no longer have the optimize argument in TF 2.0 or after (also, `tensor' argument is renamed to `input').
The calls have been converted to compat.v1 for safety (even though  they may already have been correct)..
tensorlayer/layers/recurrent.py:1051:24: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/recurrent.py:1306:24: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/recurrent.py:1581:24: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/activation.py:48:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/activation.py:120:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/activation.py:200:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/inputs.py:226:19: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/inputs.py:228:23: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/inputs.py:230:23: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/inputs.py:284:18: tf.nn.embedding_lookup requires manual check: WARNING: validate_indices argument has been removed..
tensorlayer/layers/inputs.py:290:12: tf.nn.nce_loss requires manual check: WARNING: `partition_strategy` has been removed from `tf.nn.nce_loss`  The 'div' strategy is used by default..
tensorlayer/layers/inputs.py:340:19: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/inputs.py:367:18: tf.nn.embedding_lookup requires manual check: WARNING: validate_indices argument has been removed..
tensorlayer/layers/inputs.py:413:35: tf.random_uniform_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/inputs.py:446:26: tf.nn.embedding_lookup requires manual check: WARNING: validate_indices argument has been removed..
tensorlayer/layers/utils.py:45:57: tf.zeros_like requires manual check: WARNING: tf.zeros_like and tf.ones_like no longer have the optimize argument in TF 2.0 or after (also, `tensor' argument is renamed to `input').
The calls have been converted to compat.v1 for safety (even though  they may already have been correct)..
tensorlayer/layers/utils.py:46:55: tf.zeros_like requires manual check: WARNING: tf.zeros_like and tf.ones_like no longer have the optimize argument in TF 2.0 or after (also, `tensor' argument is renamed to `input').
The calls have been converted to compat.v1 for safety (even though  they may already have been correct)..
tensorlayer/layers/utils.py:50:40: tf.ones_like requires manual check: WARNING: tf.zeros_like and tf.ones_like no longer have the optimize argument in TF 2.0 or after (also, `tensor' argument is renamed to `input').
The calls have been converted to compat.v1 for safety (even though  they may already have been correct)..
tensorlayer/layers/utils.py:51:8: tf.zeros_like requires manual check: WARNING: tf.zeros_like and tf.ones_like no longer have the optimize argument in TF 2.0 or after (also, `tensor' argument is renamed to `input').
The calls have been converted to compat.v1 for safety (even though  they may already have been correct)..
tensorlayer/layers/normalization.py:170:22: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:171:23: tf.random_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:172:29: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:202:32: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:235:28: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:304:24: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:308:73: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:452:72: tf.ones_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:453:70: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:456:83: tf.ones_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:457:81: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:506:22: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:507:23: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:508:29: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:529:79: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/normalization.py:530:77: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/scale.py:40:69: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/spatial_transformer.py:154:19: tf.ones_like requires manual check: WARNING: tf.zeros_like and tf.ones_like no longer have the optimize argument in TF 2.0 or after (also, `tensor' argument is renamed to `input').
The calls have been converted to compat.v1 for safety (even though  they may already have been correct)..
tensorlayer/layers/dense/ternary_dense.py:52:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/ternary_dense.py:53:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/quan_dense_bn.py:84:23: tf.ones_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/quan_dense_bn.py:85:22: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/quan_dense_bn.py:87:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/quan_dense_bn.py:136:58: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/quan_dense_bn.py:143:28: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/binary_dense.py:51:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/binary_dense.py:52:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/dorefa_dense.py:60:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/dorefa_dense.py:61:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/quan_dense.py:57:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/quan_dense.py:58:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/dropconnect.py:65:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/dropconnect.py:66:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/dense/dropconnect.py:108:20: WARNING: tf.nn.dropout has changed the semantics of the second argument. Please check the applied transformation.
tensorlayer/layers/convolution/ternary_conv.py:81:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/ternary_conv.py:82:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/ternary_conv.py:147:18: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/convolution/separable_conv.py:69:19: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/separable_conv.py:176:19: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_conv.py:59:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_conv.py:60:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_conv.py:189:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_conv.py:190:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_conv.py:233:18: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/convolution/expert_conv.py:292:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_conv.py:293:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/deformable_conv.py:77:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/deformable_conv.py:78:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_deconv.py:69:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_deconv.py:70:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_deconv.py:144:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_deconv.py:145:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_deconv.py:101:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_deconv.py:102:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_deconv.py:187:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/expert_deconv.py:188:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/atrous_conv.py:206:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/atrous_conv.py:207:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_conv.py:73:48: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_conv.py:74:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_conv.py:164:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/simplified_conv.py:165:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/dorefa_conv.py:73:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/dorefa_conv.py:74:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/dorefa_conv.py:140:18: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/convolution/binary_conv.py:79:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/binary_conv.py:80:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/binary_conv.py:147:18: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/convolution/quan_conv.py:85:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/quan_conv.py:86:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/quan_conv.py:152:18: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/convolution/quan_conv_bn.py:108:23: tf.ones_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/quan_conv_bn.py:109:22: tf.zeros_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/quan_conv_bn.py:113:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/quan_conv_bn.py:151:19: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/convolution/quan_conv_bn.py:174:58: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/quan_conv_bn.py:181:28: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/quan_conv_bn.py:210:24: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".
tensorlayer/layers/convolution/depthwise_conv.py:83:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/depthwise_conv.py:84:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/group_conv.py:55:19: tf.truncated_normal_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/group_conv.py:56:19: tf.constant_initializer requires manual check: WARNING: tf.initializers and tf.keras.initializers no longer have the dtype argument in the constructor or partition_info argument in the call method in TF 2.0 and after. The only API symbols are now tf.keras.initializers.* or tf.initializers.*.
The calls have been converted to compat.v1 for safety (even though they may already have been correct)..
tensorlayer/layers/convolution/group_conv.py:83:38: tf.nn.conv2d requires manual check: WARNING: use_cudnn_on_gpu argument has been removed and "filter" was renamed to "filters".

[DEKHTIARJonathan]

I would advise that you don't do anything until the API is fixed (RC released is out). The API changes all the time at the moment

[zsdonghao]

@DEKHTIARJonathan thanks, I am testing here https://github.com/zsdonghao/tensorlayer2/tree/tl2-layer

[suhailsinghbains]

Hi, Is this issue still open, I would like to work on this, I am new here.

[zsdonghao]

WELL DONE EVERYONE! https://github.com/tensorlayer/tensorlayer/releases/tag/2.0.0