Please make sure that this is a bug. As per our
GitHub Policy,
we only address code/doc bugs, performance issues, feature requests and
build/installation issues on GitHub. tag:bug_template

System information

• Have I written custom code (as opposed to using a stock
  example script provided in TensorFlow): Yes

cc @ezyang @anjali411 @dylanbespalko

What's the ETA for updating the massively outdated documentation?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

In doc.pyx' s line 590:

 if not self.is_parsed:
            raise ValueError(Errors.E029)

I can still do a good job of chunking by tokenization and pos tagging only, without the full parse. Also in some languages parse isn't available. This will leave more flexibilities to users. I can comment out this in my copy of spacy, but when I update spacy to a new release, I have to chang

Reference from TensorFlow: https://www.tensorflow.org/api_docs/cc/class/tensorflow/ops/matrix-band-part

This op is used by the Music Transformer model.

It would be beneficial to write a design document on INT8 DNNL implementation and put it into
FluidDoc next to other Design documents:
https://github.com/PaddlePaddle/FluidDoc/tree/develop/doc/fluid/design/mkldnn

Is there an overview, how the config object for the constructors of the neural network are defined?

This would be interesting:

• Has brain.NeuralNetwork and brain.recurrent.RNN the same config attributes?
• What is the description of the inputSize attribute?

Thank you for the great work with this library, it is realy amazing!

Example (from TfidfTransformer)

if isinstance(docs[0], tuple):
    docs = [docs]
return [self.gensim_model[doc] for doc in docs]

This method expects a list of tuples, instead of an iterable. This means that the entire corpus has to be stored as a lis

load_csv line can be since there's columns_to_ignore param's supported

data, labels = load_csv('titanic_dataset.csv', target_column=0, columns_to_ignore=[2, 7], categorical_labels=True, n_classes=2)

and we don't need to do that in preprocess()

def preprocess(passengers):
    for i in range(len(passengers)):
        passengers[i][1] = 1. if passengers[i][1] == 'female' else 0.

我发现examples/retinaface.cpp中，如果开启OMP加速的话似乎在检测到人脸时会发生内存泄漏，但我定位不了这个问题的具体原因。

值得注意的时，如果将qsort_descent_inplace函数中的OMP指令注释掉这个问题就会消失掉。

static void qsort_descent_inplace(std::vector<FaceObject>& faceobjects, int left, int right)
{
    int i = left;
    int j = right;
    float p = faceobjects[(left + right) / 2].prob;
    ...
    // #pragma omp parallel sections
    {
        // #pragma

Several parts of the op sec like the main op description, attributes, input and output descriptions become part of the binary that consumes ONNX e.g. onnxruntime causing an increase in its size due to strings that take no part in the execution of the model or its verification.

Setting __ONNX_NO_DOC_STRINGS doesn't really help here since (1) it's not used in the SetDoc(string) overload (s

i'm a newbie in programming. I try to use this library. it's very useful for me.
I want to show centroid in K-means clustering. how to show it? thank u so much..

I understand that these two python files show two different methods to construct a model. The original n_epoch is 500 which works perfect for both python files. But if I change n_epoch to 20, only tutorial_mnist_mlp_static.py can achieve a high test accuracy (~0.97). The other file tutorial_mnist_mlp_static_2.py only get 0.47.

The models built from these two files looks the same for me (the s

Hi I would like to propose a better implementation for 'test_indices':

We can remove the unneeded np.array casting:

Cleaner/New:
test_indices = list(set(range(len(texts))) - set(train_indices))

Old:
test_indices = np.array(list(set(range(len(texts))) - set(train_indices)))