tf.function makes invalid assumptions about arguments that are Mapping instances. In general, there are no requirements for Mapping instances to have constructors that accept [(key, value)] initializers, as assumed here.

This leads to cryptic exceptions when used with perfectly valid Mapping s

Some lines in the code block of the keras docs is too long, the result of which is, there will be a horizonal scroll bar at the bottom of the code block. That is hard to read. The long lines should be rearranged to multiple short lines to improve readibility.

Example:
The docs for the SimpleRNN class (https://keras.io/layers/recurrent/#simplernn). The initializer of SimpleRNN has m

FasterRCnn use DetectionOutput as inference results. In source code it is written

[ // num() and channels() are 1.
// Since the number of bboxes to be kept is unknown before nms, we manually
// set it to maximal number of detections, [keep_top_k] parameter multiplied by batch size.
// Each row is a 7 dimension std::vector, which stores
// [image_id, label, con

https://github.com/aymericdamien/TensorFlow-Examples/blob/master/tensorflow_v2/notebooks/3_NeuralNetworks/convolutional_network_raw.ipynb

trainable_variables = weights.values() + biases.values() doesn't work.
Also if I write trainable_variables = list(weights.values()) + list(biases.values()), I have to turn on tf.enable_eager_execution(), but the training result is wrong, accuracy is ar

Current implementation does sequential sigmoid_out and mul_. We can get better performance by fusing this operations together.

studentc@2080ti:~/caffe$ make all
PROTOC src/caffe/proto/caffe.proto
CXX .build_release/src/caffe/proto/caffe.pb.cc
CXX src/caffe/blob.cpp
CXX src/caffe/syncedmem.cpp
CXX src/caffe/net.cpp
CXX src/caffe/data_transformer.cpp
CXX src/caffe/layer.cpp
CXX src/caffe/common.cpp
CXX src/caffe/parallel.cpp
CXX src/caffe/layers/infogain_loss_layer.cpp
src/caffe/layers/infogain_loss_layer.cpp: In

Target Leakage in mentioned steps in Data Preprocessing. Train/test split needs to be before missing value imputation. Else you will have a bias in test/eval/serve.

Original line 87:
with open('README.md') as readme:

Corrected version of line 87:
with open('README.md','r',encoding='utf-8') as readme:

Explanation:
Windows uses GBK to decode rather than utf-8 at default setting

This should really help to keep a track of papers read so far. I would love to fork the repo and keep on checking the boxes in my local fork.

For example: Have a look at this section. People fork this repo and check the boxes as they finish reading each section.

Alexnet implementation in tensorflow has incomplete architecture where 2 convolution neural layers are missing. This issue is in reference to the python notebook mentioned below.

https://github.com/donnemartin/data-science-ipython-notebooks/blob/master/deep-learning/tensor-flow-examples/notebooks/3_neural_networks/alexnet.ipynb

What's the ETA for updating the massively outdated documentation?

Please update all documents that are related building CNTK from source with latest CUDA dependencies that are indicated in CNTK.Common.props and CNTK.Cpp.props.
I tried to build from source, but it's a futile effort.

I am having difficulty in running this package as a Webservice. Would appreciate if we could provide any kind of documentation on implementing an API to get the keypoints from an image. Our aim is to able to deploy this API as an Azure Function and also know if it is feasible.

I got a conllU file, from my university, where the head column is filled with .
Processing such file with the cli.convert method will result in a int cast error in
https://github.com/explosion/spaCy/blob/master/spacy/cli/converters/conllu2json.py line 73
in the read_conllx method (head = (int(head) - 1) if head != "0" else id).

In the format documentation on https://universaldependencie

With the latest version of scipy.misc, scipy.misc.toimage is no longer available. To load and save an image as png we now have to use PIL, breaking tensorboard image summary.

Here is how I fixed the bug:
1./ At the end of main.py, log a uint8 image
logger.image_summary(tag, (images * 255).astype(np.uint8), step+1)
2./ In Logger class, package image as bytes with the PIL library (mode="L

From here:

A particularity of the SV2TTS framework is that all models can be trained
separately and on distinct datasets. For the encoder, one seeks to have a model
that is robust to noise and able to capture the many characteristics of the human
voice. Therefore, a large corpus of many different speakers wou

transcribe.py has odd directory-scanning behavior which isn't documented

If you point --src to a directory, you get the error:

E Path in --src not existing

Looking at the code logic, the script expects a JSON file with a .catalog file extension. This is (1) not documented, and (2) not a really useful logic. It would be much better to point the script to a dir, and scan f