audit-trail:

• pip-10.0.1 + tweak the build process
• holoviews-1.10.1
• ipywidgets-7.2.1 , widgetsnbextension-3.2.1
• Ipython-6.3.1 and its new dependancy backcall
• think about another install path: WinPython ... to WPy-3651 (1 character less, versioning is back)
• Ipython-6.3.1
• scipy-1.1.0rc1
• further tweak the build processs: witch to https://pypi.org/project, full pip-10.0 apparently completed
• bokeh-0.12.6dev3
• Tornado-5.0.2 (require bokeh-0.12.6dev2+)
• jupyterlab-0.32.0 (beta2) : wasn't the b1 plan, but beta1 not sustainable further
• new package: loky
• numba-0.38 (python-3.7 support is only in numba-0.39)
• param-1.6.1
• cvxopt in WinPython-3.7
• pyopengl fix courtesy of cgohlke
• jupyterlab-0.32.1
• pyzo
• jupyterlab-0.32.1
• QtPy-1.4.1 patch supports PySide2-5.11.0a1-5.11.0-cp36
• patch WinPython to support Pyside2-5.11.0 and PySide2/designer.exe
• patched Spyder-3.2.8 with PyQt5 constraint removed (so we can test with PySide2)
• numpy-1.14.3
• pre-set pyzo to ipython (qtconsole), as it looks more stable with PySide2

a1 (April 29th) a build with pyside2-5.110a1 /QtPy-1.5.0dev (but QML+Tensorflow removed)

• holoviews-1.10.2 (compat with bokeh-0.12.6)
• statsmodels-0.9.0rc1
• qtlinguist icon for PySide2
• pyzo-4.5.2
• notebook-5.5.0rc1
• python-3.7.0b4
• pandas-0.23.0rc2
• altair-2.0.2
• test of PySide2-5.11.0a1.dev1525359714-5.11.0-cp36-cp36m-win_amd64: tiny progress, will wait before bothering another alpha (see Pyside to PyQt5 gap at the bottom of this page https://wiki.qt.io/Qt_for_Python_Missing_Bindings)
• experiment: binary with PyTorch instead of Legacy Tensorflow:
  • can't use recent version of Tensorflow,
  • with Tensorflow: 412875 Ko,
  • with Torch +torchvision: 409820 Ko, no error message (trying to get a fast transfer style working)
• scipy-1.1.0

b1 (may 5th) python-3.7.0b4, Tornado-5.0.2, Jupyterlab-beta2, numpy-1.14.3, early PySide2-5.11 compatibility (QtConsole, wppm, Pyzo, Qtdesigner), Altair-2, statsmodels-0.9.0rc1

• PyqtGraph-0.10.1dev2 (pyqtgraph-82afad83662b9e58d94d966b83ae3d9c00e7b6cb) (with PySide2 pull 635 current work, at may 8th)
• holoviews-1.10.3
• bokeh-0.12.6dev4 (check datashader version, may be soon)
• QtPy-1.4.2 further improves PySide2-5.11a1/a2 compatibility
• notebook-5.5.0
• vega-1.0.1 + vega3-0.13.0
• QtAwesome-0.5.0.dev0
• pandas-0.23.0 (pushed back to may 15th from April 23rd, and probably further later)
• PyQtgraph updated of pull 635 (PySide2 compat)
• numba-0.38 for python-3.7
• scikit-image-0.14.0(dev cgohlke)
• holoviews-1.10.4
• pywinpty‑0.5.3
• nodejs-8.11.2 (smaller than nodejs-8.9.4)
• traittypes-0.2.0
• statsmodels-0.9.0

b2 (may 27th, nearly feature-complete python-3.7) pandas-0.23.0, numba-0.38 on python-3.7.0b4

• python-3.7.0b5
• llvmlite-0.23.2 + numba-0.38.1 critical bug correction of Numba-0.38.0 ( but doesn't solve numba/numba#2998 )
• upgrade to holoviews-1.10.5a1( still doesn't solve numba/numba#2998 , but remove the two warnings, so we keep pandas-0.23.0)

b3 (june 3rd , python-3.7.0b5, Numba-0.38.1 critical bug correction over numba-0.38.0 )

• holoviews-0.10.5
• pyviz new stack: holoplot, parambokeh
• pandas-0.23.1 (june 8th delayed to .. ?)
• bokeh-0.13.0 ? (may postpone to june 27th if I can't get it)
• have a second look on PySide2 preview scheduled for June 8th: it doesn't crash pyzo, today. preview june 11th)
• Python-3.7.0rc1
• holoplot : the seaborn layer over holoviews
• check intake+ruamel.yaml are truly of use.
• may reconsider PyTorch vs Tensorflow choice, if pymc4 moves to Tensorflow (https://github.com/pymc-devs/pymc3/wiki/PyMC-Project-Meeting-Minutes)

b4 (june 23)

• Python-3.7.0 and Python-3.6.6
• holoviews-1.10.6
• spyder-3.3.0.dev0-20180630
• seaborn-0.9.dev0-20180629
• pyqtgraph-0.11.0.dev0-20180629
• hvplot-0.1.1
• holoviews-1.10.7b1 (links between plots)
• had a quick look at pyecharts (look nice)
• downgrade from jupyterlab-launcher-0.11.0 to jupyterlab-launcher-0.10.5 (to keep jupyterlab-0.32.1 working well)
• downgrade to jupyterlab-plugin-manager@0.35 (to keep jupyterlab-0.32.1 working well)
• numba-0.39.0/llvmlite-0.24.0 (official python-3.7 support)
• guiqwt/PyQwt and mypy for python-3.7.0
• holoviews-1.11.0a1+ (links) , datashader-0.6.7

rc (july 8th: python-3.7 final, sticking on jupyterlab-0.32.1 beta2)

• Prompt-toolkit-2.0.3/ptpython-2.0/ipython refreshed by https://github.com/ipython/ipython/pull/11177/commits
• too late for eco-system: Jupyterlab-0.33.0rc (aka beta3) (yarn-1.7 bug fix), and other js stacks
• too far away: pyside2-5.11+ compatibility, besides basic QtConsole/Pyzo.

build a1 (for alpha of PySide2 integration): 2018-04-28

This build is intended for users willing to help maturation of PySide2.

Changes from WinPython 2018-01:

• change default install directory to WPy-3651a1 (for WinPython-3.6.5.1 vs WinPython at last release)
• rework Initialization strategy to better handle PyQt5/PyQt4/PySide2, (may simplify build process)
• numpy-1.14.3, jupyterlab-0.32.1 (beta2.1)
• alpha support for PySide2-5.11.0a1-5.11.0
  . QtPy-20180429 tree
  . having to delete QML part so it fits in the installer, (uninstall then pip install it for a full experience),
  . QtConsole Works, WPPM works, Pyzo works expect crashes at closing (a PySide2 bug),
  . the rest is blowing up (or not recognizing PySide2 yet): Spyder, PyQtgraph, Rx, Guidata, ...
• adding pyzo editor, as Spyder doesn't support PySide2 yet.

Areas of interest for testers:

• does pyzo work ok ? (for users who know it well)
• participating to the great PySide2 integration in Qt-dependant packages:
  . propose patches to PyQtGraph, Spyder, Guiqwt, Guidata, Rx, ...
  . check QtDesigner works well,
• numpy-1.14.3 upgrade,

History of changes for WinPython-64bit 3.6.5.1Ps2 a1

The following changes were made to WinPython-64bit distribution since version 3.6.5.0Qt5.

Python packages

New packages:

• backcall 0.1.0 (Specifications for callback functions passed in to an API)

• geographiclib 1.49 (The geodesic routines from GeographicLib)

• loky 2.1.1 (Robust and reusable Executor for joblib)

• netcdftime 1.0.0a2 (Time-handling functionality from netcdf4-python)

• pyside2 5.11.0a1 (Python binding of the cross-platform GUI toolkit Qt)

• pyzo 4.5.1 (the Python IDE for scientific computing)

Upgraded packages:

• altair 2.0.0.dev0 → 2.0.0rc2 (High-level declarative visualization library for Python)

• astroid 1.6.2 → 1.6.3 (Rebuild a new abstract syntax tree from Python's ast (required for pylint))

• bcolz 1.2.0 → 1.2.1 (columnar and compressed data containers.)

• bokeh 0.12.15 → 0.12.16.dev3 (Statistical and novel interactive HTML plots for Python)

• brotli 1.0.2 → 1.0.4 (Python binding of the Brotli compression library)

• certifi 2018.1.18 → 2018.4.16 (Python package for providing Mozilla's CA Bundle.)

• cvxopt 1.1.9 → 1.2.0 (Convex optimization package)

• cython 0.28.1 → 0.28.2 (Cython is a language that makes writing C extensions for the Python language as easy as Python)

• decorator 4.2.1 → 4.3.0 (Better living through Python with decorators)

• distributed 1.21.4 → 1.21.6 (Distributed computing)

• fastparquet 0.1.4 → 0.1.5 (Python support for Parquet file format)

• flask 0.12.2 → 1.0 (A microframework based on Werkzeug, Jinja2 and good intentions)

• flask_cors 3.0.3 → 3.0.4 (A Flask extension adding a decorator for CORS support)

• geopy 1.12.0 → 1.13.0 (Python Geocoding Toolbox)

• h5py 2.7.1 → 2.8.0rc1 (General-purpose Python interface to HDF5 files (unlike PyTables, h5py provides direct access to the full HDF5 C library))

• holoviews 1.10.0a2 → 1.10.1 (Composable, declarative data structures for building complex visualizations easily.)

• ipyleaflet 0.7.3 → 0.8.1 (A Jupyter widget for dynamic Leaflet maps)

• ipympl 0.1.0 → 0.1.1 (Matplotlib Jupyter Extension)

• ipython 6.2.1 → 6.3.1 (Enhanced Python shell)

• ipython_sql 0.3.8 → 0.3.9 (RDBMS access via IPython)

• ipywidgets 7.2.0 → 7.2.1 (IPython HTML widgets for Jupyter)

• jedi 0.11.1 → 0.12.0 (An autocompletion tool for Python that can be used for text editors)

• jupyterlab 0.31.12 → 0.32.1 (Jupyter lab environment notebook server extension)

• keras 2.1.5 → 2.1.6 (Theano-based Deep Learning library)

• llvmlite 0.22.0 → 0.23.0 (lightweight wrapper around basic LLVM functionality)

• moviepy 0.2.3.2 → 0.2.3.4 (Video editing with Python)

• mypy 0.580 → 0.590 (Optional static typing for Python)

• netcdf4 1.3.1 → 1.3.2 (Provides an object-oriented python interface to the netCDF version 4 library)

• numba 0.37.0 → 0.38.0 (compiling Python code using LLVM)

• numcodecs 0.5.4 → 0.5.5 (buffer compression and transformation codecs for use in data storage and communication applications)

• numpy 1.13.3+mkl → 1.14.3+mkl (NumPy: multidimensional array processing for numbers, strings, records and objects (SciPy''s core module))

• numpydoc 0.7.0 → 0.8.0 (Sphinx extension to support docstrings in Numpy format)

• param 1.5.2 → 1.6.1 (Declarative Python programming using Parameters.)

• parso 0.1.1 → 0.2.0 (A Python Parser)

• pexpect 4.4.0 → 4.5.0 (Pexpect allows easy control of interactive console applications.)

• pillow 5.0.0 → 5.1.1 (Python Imaging Library (fork))

• pip 9.0.3 → 10.0.1 (A tool for installing and managing Python packages)

• psutil 5.4.3 → 5.4.5 (Provides an interface for retrieving information on all running processes and system utilization (CPU, disk, memory, network) in a portable way)

• pycodestyle 2.3.1 → 2.4.0 (Python style guide checker)

• pylint 1.8.3 → 1.8.4 (Logilab code analysis module: analyzes Python source code looking for bugs and signs of poor quality)

• pymc3 3.3 → 3.4 (Markov Chain Monte Carlo sampling toolkit.)

• pyodbc 4.0.22 → 4.0.23 (DB API Module for ODBC)

• pytest 3.5.0 → 3.5.1 (pytest: simple powerful testing with Python)

• pytz 2018.3 → 2018.4 (World Timezone Definitions for Python)

• qtpy 1.4.0 → 1.5.0.dev0 (Provides an abstraction layer on top of the various Qt bindings (PyQt5, PyQt4 and PySide) and additional custom QWidgets.)

• rope 0.10.5 → 0.10.7 (a python refactoring library...)

• rpy2 2.9.2 → 2.9.3 (Python interface to the R language (embedded R))

• scipy 1.0.1 → 1.1.0rc1 (SciPy: Scientific Library for Python (advanced math, signal processing, optimization, statistics, ...))

• setuptools 39.0.1 → 39.1.0 (Download, build, install, upgrade, and uninstall Python packages - easily)

• simplejson 3.13.2 → 3.14.0 (Simple, fast, extensible JSON (JavaScript Object Notation) encoder/decoder)

• sortedcontainers 1.5.9 → 1.5.10 (Python Sorted Container Types: SortedList, SortedDict, and SortedSet)

• sphinx 1.7.2 → 1.7.4 (Tool for generating documentation which uses reStructuredText as its markup language)

• sphinx_rtd_theme 0.2.4 → 0.3.0 (ReadTheDocs.org theme for Sphinx, 2013 version.)

• sqlalchemy 1.2.5 → 1.2.7 (SQL Toolkit and Object Relational Mapper)

• tables 3.4.2 → 3.4.3 (Package based on HDF5 library for managing hierarchical datasets (extremely large amounts of data))

• tornado 4.5.3 → 5.0.2 (Scalable, non-blocking web server and tools (required for IPython notebook))

• tqdm 4.19.9 → 4.23.1 (A Simple Python Progress Meter)

• traittypes 0.0.6 → 0.1.0 (Scipy trait types)

• vega3 0.5.0 → 0.10.0 (An IPython/ Jupyter widget for Vega 3 and Vega-Lite 2)

• wheel 0.30.0 → 0.31.0 (A built-package format for Python.)

• widgetsnbextension 3.2.0 → 3.2.1 (IPython HTML widgets for Jupyter)

• xarray 0.10.2 → 0.10.3 (N-D labeled arrays and datasets in Python)

• xlsxwriter 1.0.2 → 1.0.4 (A Python module for creating Excel XLSX files.)

Removed packages:

• absl_py 0.1.11 (Abseil Python Common Libraries)

• edward 1.3.5 (A library for probabilistic modeling, inference, and criticism. Deep generative models, variational inference. Runs on TensorFlow.)

• protobuf 3.5.2 (Protocol Buffers - Google's data interchange format)

• pyqt5 5.9.2 (Python bindings for the Qt cross platform GUI toolkit)

• sip 4.19.6 (Python extension module generator for C and C++ libraries)

• tensorflow 1.5.0 (TensorFlow helps the tensors flow)

• tensorflow_tensorboard 1.5.1 (TensorBoard lets you watch Tensors Flow)

PyTorch 0.4 is out with support for Windows.
It is really the best out there (It is also like a GPU accelerated Numpy) and I think you should include it and drop other DL to save space.

I saw PyTorch arrived, and is now slightly smaller, So I'll make a test on my favorite problem: the fast_neural_style. https://github.com/pytorch/examples/tree/master/fast_neural_style

In terms of Github popularity, it's still 10x smaller than Tensorflow, but the trend is only 40% smaller.

it's the same 40% difference I read here https://pytorch.org/2018/01/19/a-year-in.html

I read there is a tensor motor "ATen", so maybe hope for a Theano replacement.

@stonebig ,
Popularity isn't everything.
The question is which is a better tool.
PyTorch is a Python based tool while TensorFlow is only integrated with Python.

It is not like someone can not install TensorFlow on its own later.
Out of the box PyTorch should be the choice.

being only 40% behind Google after just one year of existence is a good score. can you find me a small jupyter notebook with a small "Fast Style Transfer" demo with pytorch ? (must run on a old cpu), in my memory Torch was the king of that exercise 2 years ago.

maybe this one https://github.com/zhanghang1989/PyTorch-Multi-Style-Transfer ... too big

or this, but transfer model is to find
https://github.com/alexis-jacq/Pytorch-Tutorials/blob/master/Neural_Style.ipynb

which is the official tuto
https://pytorch.org/tutorials/advanced/neural_style_tutorial.html

need something small

I see the simples tutorial uses TorchVision package, and that TorchVision will download modules needed

model_urls = {
'vgg11': 'https://download.pytorch.org/models/vgg11-bbd30ac9.pth',
'vgg13': 'https://download.pytorch.org/models/vgg13-c768596a.pth',
'vgg16': 'https://download.pytorch.org/models/vgg16-397923af.pth',
'vgg19': 'https://download.pytorch.org/models/vgg19-dcbb9e9d.pth',
'vgg11_bn': 'https://download.pytorch.org/models/vgg11_bn-6002323d.pth',
'vgg13_bn': 'https://download.pytorch.org/models/vgg13_bn-abd245e5.pth',
'vgg16_bn': 'https://download.pytorch.org/models/vgg16_bn-6c64b313.pth',
'vgg19_bn': 'https://download.pytorch.org/models/vgg19_bn-c79401a0.pth',
}

What do you mean by small?
Why does it have to be Neural Transfer?

hum ... problem: alexis-jacq/Pytorch-Tutorials#6

I'm not sure but is the optimizer LBFGS?
Try using something simpler :-).

What is wrong with trying it on MNIST + CIFAR?

yes, need you to point me on something simple , I'm going bigger and bigger: the below one model is ... 12 Go ?

https://github.com/pytorch/examples/tree/master/fast_neural_style

Pytorch:

• smaller than Tensorflow,
• latest version works on an old cpu
• may be more pythonnic, so more logical (https://agi.io/2018/02/09/survey-machine-learning-frameworks/)
• make a fast transfer style working to demonstrate this, a tutorial like latest included for Keras

The MNIST is usually the introduction task for Deep Learning:

https://github.com/pytorch/examples/blob/master/mnist/main.py

yes, but it's 300% un-sexy, to the opposite of style-transfer where you can re-paint the world in Picasso/Monet/VanGogh style.

I'm just unable to find how to save the image result of Pytorch in a file at the end of the tutorial:

• imsave doesn't work
• plt.savefig saves only blank.
  ==> an idea ?

build b1 : 2018-05-04/5

Changes from WinPython 2018-01:

• Python-3.7.0b4
• New feature: compatibility with Pyzo and PySide2
• New launchers : "Pyzo" , "QtDesigner", "QtLinguist"
• New Packages: Pytorch (replaces TensorFlow, for this build1 at least), Pyzo
• Upgraded Packages: pandas-0.23.0rc2, statsmodels-0.9.0rc1, scipy-1.1.0, numpy-1.14.3, Jupyterlab-0.32.1, altair-2.0.1....

Areas of particular interest for testers:

• Python-3.7,
• Pandas, Scipy, statsmodels, PyTorch

focus for next build:

• Pandas-0.23 final, statsmodels-0.9.0 final, bokeh-0.12.16 final,
• numba for Python-3.7.0rc1.

History of changes for WinPython-64bit 3.6.5.1Qt5 b1

The following changes were made to WinPython-64bit distribution since version 3.6.5.0Qt5

Python packages

New packages:

• backcall 0.1.0 (Specifications for callback functions passed in to an API)

• cftime 1.0.0b1 (time-handling functionality from netcdf4-python)

• geographiclib 1.49 (The geodesic routines from GeographicLib)

• loky 2.1.1 (Robust and reusable Executor for joblib)

• pyzo 4.5.2 (the Python IDE for scientific computing)

• torch 0.4.0 (a deep learning framework.)

• torchvision 0.2.1 (Datasets, Transforms and Models specific to Computer Vision)

• typing 3.6.4 (Type Hints for Python)

Upgraded packages:

• altair 2.0.0.dev0 → 2.0.1 (High-level declarative visualization library for Python)

• astroid 1.6.2 → 1.6.3 (Rebuild a new abstract syntax tree from Python's ast (required for pylint))

• attrs 17.4.0 → 18.1.0 (Classes Without Boilerplate)

• bcolz 1.2.0 → 1.2.1 (columnar and compressed data containers.)

• bokeh 0.12.15 → 0.12.16.dev3 (Statistical and novel interactive HTML plots for Python)

• brotli 1.0.2 → 1.0.4 (Python binding of the Brotli compression library)

• certifi 2018.1.18 → 2018.4.16 (Python package for providing Mozilla's CA Bundle.)

• cvxopt 1.1.9 → 1.2.0 (Convex optimization package)

• cython 0.28.1 → 0.28.2 (Cython is a language that makes writing C extensions for the Python language as easy as Python)

• dask 0.17.2 → 0.17.4 (Minimal task scheduling abstraction)

• decorator 4.2.1 → 4.3.0 (Better living through Python with decorators)

• distributed 1.21.4 → 1.21.8 (Distributed computing)

• fastparquet 0.1.4 → 0.1.5 (Python support for Parquet file format)

• flask 0.12.2 → 1.0.2 (A microframework based on Werkzeug, Jinja2 and good intentions)

• flask_cors 3.0.3 → 3.0.4 (A Flask extension adding a decorator for CORS support)

• geopy 1.12.0 → 1.13.0 (Python Geocoding Toolbox)

• h5py 2.7.1 → 2.8.0rc1 (General-purpose Python interface to HDF5 files (unlike PyTables, h5py provides direct access to the full HDF5 C library))

• holoviews 1.10.0a2 → 1.10.2 (Composable, declarative data structures for building complex visualizations easily.)

• ipyleaflet 0.7.3 → 0.8.1 (A Jupyter widget for dynamic Leaflet maps)

• ipympl 0.1.0 → 0.1.1 (Matplotlib Jupyter Extension)

• ipython 6.2.1 → 6.3.1 (Enhanced Python shell)

• ipython_sql 0.3.8 → 0.3.9 (RDBMS access via IPython)

• ipywidgets 7.2.0 → 7.2.1 (IPython HTML widgets for Jupyter)

• jedi 0.11.1 → 0.12.0 (An autocompletion tool for Python that can be used for text editors)

• jupyterlab 0.31.12 → 0.32.1 (Jupyter lab environment notebook server extension)

• keras 2.1.5 → 2.1.6 (Theano-based Deep Learning library)

• llvmlite 0.22.0 → 0.23.0 (lightweight wrapper around basic LLVM functionality)

• monotonic 1.4 → 1.5 (An implementation of time.monotonic() for Python 2 & < 3.3)

• moviepy 0.2.3.2 → 0.2.3.4 (Video editing with Python)

• mypy 0.580 → 0.600 (Optional static typing for Python)

• netcdf4 1.3.1 → 1.4.0 (Provides an object-oriented python interface to the netCDF version 4 library)

• notebook 5.4.1 → 5.5.0rc1 (# Jupyter Notebook)

• numba 0.37.0 → 0.38.0 (compiling Python code using LLVM)

• numcodecs 0.5.4 → 0.5.5 (buffer compression and transformation codecs for use in data storage and communication applications)

• numexpr 2.6.4 → 2.6.5 (Fast evaluation of array expressions elementwise by using a vector-based virtual machine)

• numpy 1.13.3+mkl → 1.14.3+mkl (NumPy: multidimensional array processing for numbers, strings, records and objects (SciPy''s core module))

• numpydoc 0.7.0 → 0.8.0 (Sphinx extension to support docstrings in Numpy format)

• pandas 0.22.0 → 0.23.0rc2 (Powerful data structures for data analysis, time series and statistics)

• param 1.5.2 → 1.6.1 (Declarative Python programming using Parameters.)

• parso 0.1.1 → 0.2.0 (A Python Parser)

• pexpect 4.4.0 → 4.5.0 (Pexpect allows easy control of interactive console applications.)

• pillow 5.0.0 → 5.1.1 (Python Imaging Library (fork))

• pip 9.0.3 → 10.0.1 (A tool for installing and managing Python packages)

• psutil 5.4.3 → 5.4.5 (Provides an interface for retrieving information on all running processes and system utilization (CPU, disk, memory, network) in a portable way)

• pybind11 2.2.2 → 2.2.3 (Seamless operability between C++11 and Python)

• pycodestyle 2.3.1 → 2.4.0 (Python style guide checker)

• pylint 1.8.3 → 1.8.4 (Logilab code analysis module: analyzes Python source code looking for bugs and signs of poor quality)

• pymc3 3.3 → 3.4.1 (Markov Chain Monte Carlo sampling toolkit.)

• pyodbc 4.0.22 → 4.0.23 (DB API Module for ODBC)

• pytest 3.5.0 → 3.5.1 (pytest: simple powerful testing with Python)

• pytz 2018.3 → 2018.4 (World Timezone Definitions for Python)

• qtpy 1.4.0 → 1.5.0.dev0 (Provides an abstraction layer on top of the various Qt bindings (PyQt5, PyQt4 and PySide) and additional custom QWidgets.)

• rope 0.10.5 → 0.10.7 (a python refactoring library...)

• rpy2 2.9.2 → 2.9.3 (Python interface to the R language (embedded R))

• scipy 1.0.1 → 1.1.0 (SciPy: Scientific Library for Python (advanced math, signal processing, optimization, statistics, ...))

• setuptools 39.0.1 → 39.1.0 (Download, build, install, upgrade, and uninstall Python packages - easily)

• simplejson 3.13.2 → 3.14.0 (Simple, fast, extensible JSON (JavaScript Object Notation) encoder/decoder)

• sortedcontainers 1.5.9 → 1.5.10 (Python Sorted Container Types: SortedList, SortedDict, and SortedSet)

• sphinx 1.7.2 → 1.7.4 (Tool for generating documentation which uses reStructuredText as its markup language)

• sphinx_rtd_theme 0.2.4 → 0.3.0 (ReadTheDocs.org theme for Sphinx, 2013 version.)

• sqlalchemy 1.2.5 → 1.2.7 (SQL Toolkit and Object Relational Mapper)

• statsmodels 0.9.0.dev2 → 0.9.0rc1 (Statistical computations and models for use with SciPy)

• tables 3.4.2 → 3.4.3 (Package based on HDF5 library for managing hierarchical datasets (extremely large amounts of data))

• tornado 4.5.3 → 5.0.2 (Scalable, non-blocking web server and tools (required for IPython notebook))

• tqdm 4.19.9 → 4.23.2 (A Simple Python Progress Meter)

• traittypes 0.0.6 → 0.1.0 (Scipy trait types)

• vega3 0.5.0 → 0.11.0 (An IPython/ Jupyter widget for Vega 3 and Vega-Lite 2)

• wheel 0.30.0 → 0.31.0 (A built-package format for Python.)

• widgetsnbextension 3.2.0 → 3.2.1 (IPython HTML widgets for Jupyter)

• xarray 0.10.2 → 0.10.3 (N-D labeled arrays and datasets in Python)

• xlsxwriter 1.0.2 → 1.0.4 (A Python module for creating Excel XLSX files.)

Removed packages:

• absl_py 0.1.11 (Abseil Python Common Libraries)

• msgpack_python 0.5.4+dummy (MessagePack (de)serializer.)

• protobuf 3.5.2 (Protocol Buffers - Google's data interchange format)

• tensorflow 1.5.0 (TensorFlow helps the tensors flow)

• tensorflow_tensorboard 1.5.1 (TensorBoard lets you watch Tensors Flow)

Well,
If the data is stored as a PytorchTensor just do the following:

mA = mT.cpu().numpy()

This will convert the tensor mT to Numpy array mA.
From there, just do as you usually do with Numpy arrays.

P. S.
In PyTorch 0.4.0 they suggest using the .device() and .to() functionality.
So it would be mA = mT.to("cpu") (See PyTorch 0.4.0 Migration Guide at the chapter Writing Device Agnostic Code).

build b2: 2018-05-27 (may 27th, nearly feature-complete python-3.7)

Changes from b1:

• New feature: numba-0.38.0 on Python-3.7.0.b4, and dependant packages
• Upgraded Packages: pandas-0.23.0, statsmodels-0.9.0, pywinpty-0.5.3, Holoviews-1.10.4, bokeh-0.12.16
• Upgraded Tools: npm-8.11.2

Areas of particular interest for testers:

• Python-3.7 nearly complete stack (missing: pythonnet+clrmagic, cyordereddict, guiqwt, pyarrow, feather_format, mypy). Winpython_checker works in Jupyterlab for all but (missing) pythonnet ("import clr")
• a doubt on Holoviews/datashader/outliers test problem on Python-3.7 64bit (32bit is ok) (ioam/holoviews#2738)

focus for next build:

• Python-3.7.0rc1,

History of changes for WinPython-64bit 3.7.0.1 b2

The following changes were made to WinPython-64bit distribution since version 3.7.0.0.

Tools

Upgraded packages:

• Nodejs v8.9.4 → v8.11.2 (a JavaScript runtime built on Chrome's V8 JavaScript engine)

Python packages

New packages:

• atomicwrites 1.1.5 (Powerful Python library for atomic file writes.)

• backcall 0.1.0 (Specifications for callback functions passed in to an API)

• cftime 1.0.0 (time-handling functionality from netcdf4-python)

• colorcet 0.9.1 (A set of useful perceptually uniform colormaps for plotting scientific data)

• cvxopt 1.2.0 (Convex optimization package)

• cvxpy 1.0.3 (A domain-specific language for modeling convex optimization problems in Python)

• datashader 0.6.6 (Data visualization toolchain based on aggregating into a grid)

• dill 0.2.7.1 (serialize all of python (almost))

• ecos 2.0.5 (This is the Python package for ECOS: Embedded Cone Solver)

• fastcache 1.0.2 (C implementation of Python 3 functools.lru_cache)

• fastparquet 0.1.5 (Python support for Parquet file format)

• future 0.16.0 (Clean single-source support for Python 3 and 2)

• geographiclib 1.49 (The geodesic routines from GeographicLib)

• gmpy2 2.0.8 (GMP/MPIR, MPFR, and MPC interface to Python 2.6+ and 3.x)

• llvmlite 0.23.1 (lightweight wrapper around basic LLVM functionality)

• loky 2.1.1 (Robust and reusable Executor for joblib)

• multiprocess 0.70.5 (better multiprocessing and multithreading in python)

• numba 0.38.0 (compiling Python code using LLVM)

• osqp 0.3.0 (the Operator Splitting QP Solver.)

• pyct 0.4.2 ()

• pyzo 4.5.2 (the Python IDE for scientific computing)

• scs 1.2.7 (scs: splitting conic solver)

• thrift 0.11.0 (a software framework for scalable cross-language services development)

• torch 0.4.0 (a deep learning framework.)

• torchvision 0.2.1 (Datasets, Transforms and Models specific to Computer Vision)

• typing 3.6.4 (Type Hints for Python)

• uncertainties 3.0.2 (Transparent calculations with uncertainties on the quantities involved (aka error propagation); fast calculation of derivatives)

Upgraded packages:

• altair 2.0.0.dev0 → 2.0.1 (High-level declarative visualization library for Python)

• astroid 1.6.2 → 1.6.4 (Rebuild a new abstract syntax tree from Python's ast (required for pylint))

• attrs 17.4.0 → 18.1.0 (Classes Without Boilerplate)

• bcolz 1.2.0 → 1.2.1 (columnar and compressed data containers.)

• bokeh 0.12.15 → 0.12.16 (Statistical and novel interactive HTML plots for Python)

• brotli 1.0.2 → 1.0.4 (Python binding of the Brotli compression library)

• certifi 2018.1.18 → 2018.4.16 (Python package for providing Mozilla's CA Bundle.)

• cloudpickle 0.5.2 → 0.5.3 (Extended pickling support for Python objects)

• cython 0.28.1 → 0.28.2 (Cython is a language that makes writing C extensions for the Python language as easy as Python)

• dask 0.17.2 → 0.17.5 (Minimal task scheduling abstraction)

• decorator 4.2.1 → 4.3.0 (Better living through Python with decorators)

• distributed 1.21.4 → 1.21.8 (Distributed computing)

• flask 0.12.2 → 1.0.2 (A microframework based on Werkzeug, Jinja2 and good intentions)

• flask_cors 3.0.3 → 3.0.4 (A Flask extension adding a decorator for CORS support)

• geopy 1.12.0 → 1.14.0 (Python Geocoding Toolbox)

• h5py 2.8.0rc1 → 2.8.0 (General-purpose Python interface to HDF5 files (unlike PyTables, h5py provides direct access to the full HDF5 C library))

• holoviews 1.10.0a2 → 1.10.4 (Composable, declarative data structures for building complex visualizations easily.)

• ipyleaflet 0.7.3 → 0.8.1 (A Jupyter widget for dynamic Leaflet maps)

• ipympl 0.1.0 → 0.1.1 (Matplotlib Jupyter Extension)

• ipython 6.2.1 → 6.4.0 (Enhanced Python shell)

• ipython_sql 0.3.8 → 0.3.9 (RDBMS access via IPython)

• ipywidgets 7.2.0 → 7.2.1 (IPython HTML widgets for Jupyter)

• jedi 0.11.1 → 0.12.0 (An autocompletion tool for Python that can be used for text editors)

• jupyterlab 0.31.12 → 0.32.1 (Jupyter lab environment notebook server extension)

• keras 2.1.5 → 2.1.6 (Theano-based Deep Learning library)

• lmfit 0.9.9 → 0.9.10 (Least-Squares Minimization with Bounds and Constraints)

• monotonic 1.4 → 1.5 (An implementation of time.monotonic() for Python 2 & < 3.3)

• moviepy 0.2.3.2 → 0.2.3.4 (Video editing with Python)

• netcdf4 1.3.1 → 1.4.0 (Provides an object-oriented python interface to the netCDF version 4 library)

• nltk 3.2.5 → 3.3 (The Natural Language Toolkit (NLTK) is a Python package for natural language processing.)

• notebook 5.4.1 → 5.5.0 (# Jupyter Notebook)

• numcodecs 0.5.4 → 0.5.5 (buffer compression and transformation codecs for use in data storage and communication applications)

• numexpr 2.6.4 → 2.6.5 (Fast evaluation of array expressions elementwise by using a vector-based virtual machine)

• numpy 1.14.2+mkl → 1.14.3+mkl (NumPy: multidimensional array processing for numbers, strings, records and objects (SciPy''s core module))

• numpydoc 0.7.0 → 0.8.0 (Sphinx extension to support docstrings in Numpy format)

• palettable 3.1.0 → 3.1.1 (Color palettes for Python)

• pandas 0.22.0 → 0.23.0 (Powerful data structures for data analysis, time series and statistics)

• param 1.5.2 → 1.6.1 (Declarative Python programming using Parameters.)

• parso 0.1.1 → 0.2.1 (A Python Parser)

• pexpect 4.4.0 → 4.5.0 (Pexpect allows easy control of interactive console applications.)

• pillow 5.0.0 → 5.1.1 (Python Imaging Library (fork))

• pip 9.0.3 → 10.0.1 (A tool for installing and managing Python packages)

• psutil 5.4.3 → 5.4.5 (Provides an interface for retrieving information on all running processes and system utilization (CPU, disk, memory, network) in a portable way)

• pybind11 2.2.2 → 2.2.3 (Seamless operability between C++11 and Python)

• pycodestyle 2.3.1 → 2.4.0 (Python style guide checker)

• pyflakes 1.6.0 → 2.0.0 (passive checker of Python programs)

• pylint 1.8.3 → 1.9.1 (Logilab code analysis module: analyzes Python source code looking for bugs and signs of poor quality)

• pymc3 3.3 → 3.4.1 (Markov Chain Monte Carlo sampling toolkit.)

• pyqtgraph 0.10.1.dev1 → 0.10.1.dev2 (Scientific Graphics and GUI Library for Python)

• pytest 3.5.0 → 3.6.0 (pytest: simple powerful testing with Python)

• python_dateutil 2.7.2 → 2.7.3 (Powerful extensions to the standard datetime module)

• pytz 2018.3 → 2018.4 (World Timezone Definitions for Python)

• pywinpty 0.5.1 → 0.5.3 (Python bindings for the winpty library)

• qtawesome 0.4.4 → 0.5.0.dev0 (FontAwesome icons in PyQt and PySide applications)

• qtpy 1.4.0 → 1.4.2 (Provides an abstraction layer on top of the various Qt bindings (PyQt5, PyQt4 and PySide) and additional custom QWidgets.)

• rope 0.10.5 → 0.10.7 (a python refactoring library...)

• rpy2 2.9.2 → 2.9.3 (Python interface to the R language (embedded R))

• scikit_image 0.13.1 → 0.14.0 (Image processing toolbox for SciPy)

• scipy 1.0.1 → 1.1.0 (SciPy: Scientific Library for Python (advanced math, signal processing, optimization, statistics, ...))

• setuptools 39.0.1 → 39.2.0 (Download, build, install, upgrade, and uninstall Python packages - easily)

• simplejson 3.13.2 → 3.14.0 (Simple, fast, extensible JSON (JavaScript Object Notation) encoder/decoder)

• sortedcontainers 1.5.9 → 2.0.2 (Python Sorted Container Types: SortedList, SortedDict, and SortedSet)

• sounddevice 0.3.10 → 0.3.11 (Play and Record Sound with Python)

• sphinx 1.7.2 → 1.7.4 (Tool for generating documentation which uses reStructuredText as its markup language)

• sphinx_rtd_theme 0.2.4 → 0.3.1 (ReadTheDocs.org theme for Sphinx, 2013 version.)

• sqlalchemy 1.2.6 → 1.2.7 (SQL Toolkit and Object Relational Mapper)

• statsmodels 0.8.0 → 0.9.0 (Statistical computations and models for use with SciPy)

• tables 3.4.2 → 3.4.3 (Package based on HDF5 library for managing hierarchical datasets (extremely large amounts of data))

• theano 1.0.1 → 1.0.2 (Optimizing compiler for evaluating mathematical expressions on CPUs and GPUs.)

• tornado 4.5.3 → 5.0.2 (Scalable, non-blocking web server and tools (required for IPython notebook))

• tqdm 4.19.9 → 4.23.4 (A Simple Python Progress Meter)

• traittypes 0.0.6 → 0.2.0 (Scipy trait types)

• vega 0.5.0 → 1.1.1 (An IPython/ Jupyter widget for Vega and Vega-Lite)

• vega3 0.5.0 → 0.13.0 (An IPython/ Jupyter widget for Vega 3 and Vega-Lite 2)

• wheel 0.30.0 → 0.31.1 (A built-package format for Python.)

• widgetsnbextension 3.2.0 → 3.2.1 (IPython HTML widgets for Jupyter)

• xarray 0.10.2 → 0.10.4 (N-D labeled arrays and datasets in Python)

• xlsxwriter 1.0.2 → 1.0.5 (A Python module for creating Excel XLSX files.)

Removed packages:

• absl_py 0.1.11 (Abseil Python Common Libraries)

• astor 0.6.2 (Read/rewrite/write Python ASTs)

• gast 0.2.0 (Python AST that abstracts the underlying Python version)

• grpcio 1.10.0 (HTTP/2-based RPC framework)

• msgpack_python 0.5.4+dummy (MessagePack (de)serializer.)

• protobuf 3.5.1 (Protocol Buffers - Google's data interchange format)

• tensorboard 1.7.0 (TensorBoard lets you watch Tensors Flow)

• tensorflow 1.7.0 (TensorFlow helps the tensors flow)

• termcolor 1.1.0 (ANSII Color formatting for output in terminal)

• vega_datasets 0.4.1 (A Python package for offline access to Vega datasets)

build b3: 2018-06-02/3 (python-3.7.0b5 , as python-3.7.0rc is delayed)

Changes from b2:

• debugged Python-3.7.0b5 ; numba-0.38.1, holoviews-0.10.5a1
• updated numeric packages,

Areas of particular interest for testers:

• Python-3.7 should be fully stable now, some decrepited package may not make it . Winpython_checker works in Jupyterlab for all but (missing) pythonnet ("import clr")
• re-fresh of numeric packages: numpy, scipy, scikit-image

focus for next build b4 (next week-end):

• holoviews-1.10.5
• pandas-0.23.1 (uncertain),
• bokeh-0.13.0 (unlikely)
• maybe an effort on PySide2 preview RC, if it doesn't crash on pyzo exit
• rc around june 24th

History of changes for WinPython-64bit 3.6.5.1Qt5 b3

The following changes were made to WinPython-64bit distribution since version 3.6.5.0Qt5.

Tools

Upgraded packages:

• Nodejs v8.9.4 → v8.11.2 (a JavaScript runtime built on Chrome's V8 JavaScript engine)

Python packages

New packages:

• atomicwrites 1.1.5 (Powerful Python library for atomic file writes.)

• backcall 0.1.0 (Specifications for callback functions passed in to an API)

• cftime 1.0.0 (time-handling functionality from netcdf4-python)

• future 0.16.0 (Clean single-source support for Python 3 and 2)

• geographiclib 1.49 (The geodesic routines from GeographicLib)

• loky 2.1.1 (Robust and reusable Executor for joblib)

• osqp 0.3.0 (the Operator Splitting QP Solver.)

• pyct 0.4.2 ()

• pyzo 4.5.2 (the Python IDE for scientific computing)

• torch 0.4.0 (a deep learning framework.)

• torchvision 0.2.1 (Datasets, Transforms and Models specific to Computer Vision)

• typing 3.6.4 (Type Hints for Python)

• uncertainties 3.0.2 (Transparent calculations with uncertainties on the quantities involved (aka error propagation); fast calculation of derivatives)

Upgraded packages:

• altair 2.0.0.dev0 → 2.0.1 (High-level declarative visualization library for Python)

• astroid 1.6.2 → 1.6.4 (Rebuild a new abstract syntax tree from Python's ast (required for pylint))

• attrs 17.4.0 → 18.1.0 (Classes Without Boilerplate)

• babel 2.5.3 → 2.6.0 (Internationalization utilities)

• bcolz 1.2.0 → 1.2.1 (columnar and compressed data containers.)

• bokeh 0.12.15 → 0.12.16 (Statistical and novel interactive HTML plots for Python)

• brotli 1.0.2 → 1.0.4 (Python binding of the Brotli compression library)

• certifi 2018.1.18 → 2018.4.16 (Python package for providing Mozilla's CA Bundle.)

• cloudpickle 0.5.2 → 0.5.3 (Extended pickling support for Python objects)

• cvxopt 1.1.9 → 1.2.0 (Convex optimization package)

• cvxpy 0.4.11 → 1.0.3 (A domain-specific language for modeling convex optimization problems in Python)

• cython 0.28.1 → 0.28.3 (Cython is a language that makes writing C extensions for the Python language as easy as Python)

• dask 0.17.2 → 0.17.5 (Minimal task scheduling abstraction)

• datashader 0.6.5 → 0.6.6 (Data visualization toolchain based on aggregating into a grid)

• decorator 4.2.1 → 4.3.0 (Better living through Python with decorators)

• distributed 1.21.4 → 1.21.8 (Distributed computing)

• fastparquet 0.1.4 → 0.1.5 (Python support for Parquet file format)

• flask 0.12.2 → 1.0.2 (A microframework based on Werkzeug, Jinja2 and good intentions)

• flask_cors 3.0.3 → 3.0.4 (A Flask extension adding a decorator for CORS support)

• geopy 1.12.0 → 1.14.0 (Python Geocoding Toolbox)

• h5py 2.7.1 → 2.8.0 (General-purpose Python interface to HDF5 files (unlike PyTables, h5py provides direct access to the full HDF5 C library))

• holoviews 1.10.0a2 → 1.10.5a1 (Composable, declarative data structures for building complex visualizations easily.)

• ipydatawidgets 3.0.0 → 3.1.0 (A set of widgets to help facilitate reuse of large datasets across widgets)

• ipyleaflet 0.7.3 → 0.8.1 (A Jupyter widget for dynamic Leaflet maps)

• ipympl 0.1.0 → 0.1.1 (Matplotlib Jupyter Extension)

• ipython 6.2.1 → 6.4.0 (Enhanced Python shell)

• ipython_sql 0.3.8 → 0.3.9 (RDBMS access via IPython)

• ipywidgets 7.2.0 → 7.2.1 (IPython HTML widgets for Jupyter)

• jedi 0.11.1 → 0.12.0 (An autocompletion tool for Python that can be used for text editors)

• jupyterlab 0.31.12 → 0.32.1 (Jupyter lab environment notebook server extension)

• keras 2.1.5 → 2.1.6 (Theano-based Deep Learning library)

• llvmlite 0.22.0 → 0.23.2 (lightweight wrapper around basic LLVM functionality)

• lmfit 0.9.9 → 0.9.10 (Least-Squares Minimization with Bounds and Constraints)

• monotonic 1.4 → 1.5 (An implementation of time.monotonic() for Python 2 & < 3.3)

• more_itertools 4.1.0 → 4.2.0 (More routines for operating on iterables, beyond itertools)

• moviepy 0.2.3.2 → 0.2.3.5 (Video editing with Python)

• mypy 0.580 → 0.600 (Optional static typing for Python)

• netcdf4 1.3.1 → 1.4.0 (Provides an object-oriented python interface to the netCDF version 4 library)

• nltk 3.2.5 → 3.3 (The Natural Language Toolkit (NLTK) is a Python package for natural language processing.)

• notebook 5.4.1 → 5.5.0 (# Jupyter Notebook)

• numba 0.37.0 → 0.38.1 (compiling Python code using LLVM)

• numcodecs 0.5.4 → 0.5.5 (buffer compression and transformation codecs for use in data storage and communication applications)

• numexpr 2.6.4 → 2.6.5 (Fast evaluation of array expressions elementwise by using a vector-based virtual machine)

• numpy 1.13.3+mkl → 1.14.3+mkl (NumPy: multidimensional array processing for numbers, strings, records and objects (SciPy''s core module))

• numpydoc 0.7.0 → 0.8.0 (Sphinx extension to support docstrings in Numpy format)

• palettable 3.1.0 → 3.1.1 (Color palettes for Python)

• pandas 0.22.0 → 0.23.0 (Powerful data structures for data analysis, time series and statistics)

• param 1.5.2 → 1.6.1 (Declarative Python programming using Parameters.)

• parso 0.1.1 → 0.2.1 (A Python Parser)

• pexpect 4.4.0 → 4.6.0+dummy (Pexpect allows easy control of interactive console applications.)

• pillow 5.0.0 → 5.1.1 (Python Imaging Library (fork))

• pip 9.0.3 → 10.0.1 (A tool for installing and managing Python packages)

• psutil 5.4.3 → 5.4.5 (Provides an interface for retrieving information on all running processes and system utilization (CPU, disk, memory, network) in a portable way)

• pybind11 2.2.2 → 2.2.3 (Seamless operability between C++11 and Python)

• pycodestyle 2.3.1 → 2.4.0 (Python style guide checker)

• pyflakes 1.6.0 → 2.0.0 (passive checker of Python programs)

• pylint 1.8.3 → 1.9.1 (Logilab code analysis module: analyzes Python source code looking for bugs and signs of poor quality)

• pymc3 3.3 → 3.4.1 (Markov Chain Monte Carlo sampling toolkit.)

• pyodbc 4.0.22 → 4.0.23 (DB API Module for ODBC)

• pyqtgraph 0.10.1.dev1 → 0.10.1.dev2 (Scientific Graphics and GUI Library for Python)

• pytest 3.5.0 → 3.6.0 (pytest: simple powerful testing with Python)

• python_dateutil 2.7.2 → 2.7.3 (Powerful extensions to the standard datetime module)

• pytz 2018.3 → 2018.4 (World Timezone Definitions for Python)

• pywinpty 0.5.1 → 0.5.3 (Python bindings for the winpty library)

• qtawesome 0.4.4 → 0.5.0.dev0 (FontAwesome icons in PyQt and PySide applications)

• qtpy 1.4.0 → 1.4.2 (Provides an abstraction layer on top of the various Qt bindings (PyQt5, PyQt4 and PySide) and additional custom QWidgets.)

• rope 0.10.5 → 0.10.7 (a python refactoring library...)

• rpy2 2.9.2 → 2.9.3 (Python interface to the R language (embedded R))

• scikit_image 0.13.1 → 0.14.0 (Image processing toolbox for SciPy)

• scipy 1.0.1 → 1.1.0 (SciPy: Scientific Library for Python (advanced math, signal processing, optimization, statistics, ...))

• setuptools 39.0.1 → 39.2.0 (Download, build, install, upgrade, and uninstall Python packages - easily)

• simplejson 3.13.2 → 3.14.0 (Simple, fast, extensible JSON (JavaScript Object Notation) encoder/decoder)

• sortedcontainers 1.5.9 → 2.0.2 (Python Sorted Container Types: SortedList, SortedDict, and SortedSet)

• sounddevice 0.3.10 → 0.3.11 (Play and Record Sound with Python)

• sphinx 1.7.2 → 1.7.5 (Tool for generating documentation which uses reStructuredText as its markup language)

• sphinx_rtd_theme 0.2.4 → 0.3.1 (ReadTheDocs.org theme for Sphinx, 2013 version.)

• sqlalchemy 1.2.5 → 1.2.8 (SQL Toolkit and Object Relational Mapper)

• statsmodels 0.9.0.dev2 → 0.9.0 (Statistical computations and models for use with SciPy)

• tables 3.4.2 → 3.4.3 (Package based on HDF5 library for managing hierarchical datasets (extremely large amounts of data))

• theano 1.0.1 → 1.0.2 (Optimizing compiler for evaluating mathematical expressions on CPUs and GPUs.)

• tornado 4.5.3 → 5.0.2 (Scalable, non-blocking web server and tools (required for IPython notebook))

• tqdm 4.19.9 → 4.23.4 (A Simple Python Progress Meter)

• traittypes 0.0.6 → 0.2.0 (Scipy trait types)

• vega 0.5.0 → 1.1.1 (An IPython/ Jupyter widget for Vega and Vega-Lite)

• vega3 0.5.0 → 0.13.0 (An IPython/ Jupyter widget for Vega 3 and Vega-Lite 2)

• vega_datasets 0.4.1 → 0.5.0 (A Python package for offline access to Vega datasets)

• wheel 0.30.0 → 0.31.1 (A built-package format for Python.)

• widgetsnbextension 3.2.0 → 3.2.1 (IPython HTML widgets for Jupyter)

• xarray 0.10.2 → 0.10.4 (N-D labeled arrays and datasets in Python)

• xlsxwriter 1.0.2 → 1.0.5 (A Python module for creating Excel XLSX files.)

Removed packages:

• absl_py 0.1.11 (Abseil Python Common Libraries)

• cvxcanon 0.1.1 (common operations for convex optimization modeling tools.)

• msgpack_python 0.5.4+dummy (MessagePack (de)serializer.)

• protobuf 3.5.2 (Protocol Buffers - Google's data interchange format)

• ptyprocess 0.5.2 (Run a subprocess in a pseudo terminal)

• tensorflow 1.5.0 (TensorFlow helps the tensors flow)

• tensorflow_tensorboard 1.5.1 (TensorBoard lets you watch Tensors Flow)

build b5: 2018-06-23/24 (python-3.7.0rc1 )

Changes from b3:

• Python-3.7.0rc1,
• Bokeh-0.13.0, holoviews-0.10.5, holoplot, ipyleaflet-0.8.4
• Pandas-0.23.1

Areas of particular interest for testers:

• holoviews-1.10.5, pandas-0.23.1 , bokeh-0.13.0
• ipyleaflet

focus for next build (rc next week-end ?):

• python-3.7.0 final , python-3.6.6
• spyder-3.3.0 (with separation of spyder-kernels)

History of changes for WinPython-64bit 3.7.0.1

The following changes were made to WinPython-64bit distribution since version 3.7.0.0.

Tools

Upgraded packages:

• Nodejs v8.9.4 → v8.11.2 (a JavaScript runtime built on Chrome's V8 JavaScript engine)

Python packages

New packages:

• appdirs 1.4.3 (A small Python module for determining appropriate " + "platform-specific dirs, e.g. a "user data dir".)

• atomicwrites 1.1.5 (Powerful Python library for atomic file writes.)

• backcall 0.1.0 (Specifications for callback functions passed in to an API)

• cftime 1.0.0 (time-handling functionality from netcdf4-python)

• colorcet 0.9.1 (A set of useful perceptually uniform colormaps for plotting scientific data)

• cvxopt 1.2.0 (Convex optimization package)

• cvxpy 1.0.6 (A domain-specific language for modeling convex optimization problems in Python)

• datashader 0.6.6 (Data visualization toolchain based on aggregating into a grid)

• dill 0.2.8.2 (serialize all of python (almost))

• ecos 2.0.5 (This is the Python package for ECOS: Embedded Cone Solver)

• fastcache 1.0.2 (C implementation of Python 3 functools.lru_cache)

• fastparquet 0.1.5 (Python support for Parquet file format)

• future 0.16.0 (Clean single-source support for Python 3 and 2)

• geographiclib 1.49 (The geodesic routines from GeographicLib)

• gmpy2 2.0.8 (GMP/MPIR, MPFR, and MPC interface to Python 2.6+ and 3.x)

• holoplot 0.1.0 (A high-level plotting API for pandas, dask, streamz and xarray built on HoloViews)

• intake 0.1.3 (Data input plugin and catalog system)

• keras_applications 1.0.2 ()

• keras_preprocessing 1.0.1 ()

• llvmlite 0.23.2 (lightweight wrapper around basic LLVM functionality)

• loky 2.1.1 (Robust and reusable Executor for joblib)

• msgpack_numpy 0.4.3 ()

• multiprocess 0.70.6.1 (better multiprocessing and multithreading in python)

• numba 0.38.1 (compiling Python code using LLVM)

• osqp 0.3.0 (the Operator Splitting QP Solver.)

• parambokeh 0.2.2 (Declarative Python programming using Parameters.)

• pyct 0.4.2 ()

• pyviz_comms 0.1.1 (Launch jobs, organize the output, and dissect the results)

• pyzo 4.5.2 (the Python IDE for scientific computing)

• ruamel.yaml 0.15.40 ()

• scs 1.2.7 (scs: splitting conic solver)

• thrift 0.11.0 (a software framework for scalable cross-language services development)

• torch 0.4.0 (a deep learning framework.)

• torchvision 0.2.1 (Datasets, Transforms and Models specific to Computer Vision)

• typing 3.6.4 (Type Hints for Python)

• uncertainties 3.0.2 (Transparent calculations with uncertainties on the quantities involved (aka error propagation); fast calculation of derivatives)

Upgraded packages:

• altair 2.0.0.dev0 → 2.1.0 (High-level declarative visualization library for Python)

• astroid 1.6.2 → 1.6.5 (Rebuild a new abstract syntax tree from Python's ast (required for pylint))

• attrs 17.4.0 → 18.1.0 (Classes Without Boilerplate)

• babel 2.5.3 → 2.6.0 (Internationalization utilities)

• bcolz 1.2.0 → 1.2.1 (columnar and compressed data containers.)

• bokeh 0.12.15 → 0.13.0 (Statistical and novel interactive HTML plots for Python)

• brotli 1.0.2 → 1.0.4 (Python binding of the Brotli compression library)

• certifi 2018.1.18 → 2018.4.16 (Python package for providing Mozilla's CA Bundle.)

• cloudpickle 0.5.2 → 0.5.3 (Extended pickling support for Python objects)

• cython 0.28.1 → 0.28.3 (Cython is a language that makes writing C extensions for the Python language as easy as Python)

• dask 0.17.2 → 0.18.1 (Minimal task scheduling abstraction)

• decorator 4.2.1 → 4.3.0 (Better living through Python with decorators)

• distributed 1.21.4 → 1.22.0 (Distributed computing)

• flask 0.12.2 → 1.0.2 (A microframework based on Werkzeug, Jinja2 and good intentions)

• flask_cors 3.0.3 → 3.0.6 (A Flask extension adding a decorator for CORS support)

• geopy 1.12.0 → 1.14.0 (Python Geocoding Toolbox)

• h5py 2.8.0rc1 → 2.8.0 (General-purpose Python interface to HDF5 files (unlike PyTables, h5py provides direct access to the full HDF5 C library))

• holoviews 1.10.0a2 → 1.10.5 (Composable, declarative data structures for building complex visualizations easily.)

• idna 2.6 → 2.7 (Internationalized Domain Names in Applications (IDNA))

• ipydatawidgets 3.0.0 → 3.1.0 (A set of widgets to help facilitate reuse of large datasets across widgets)

• ipyleaflet 0.7.3 → 0.8.4 (A Jupyter widget for dynamic Leaflet maps)

• ipympl 0.1.0 → 0.1.1 (Matplotlib Jupyter Extension)

• ipyparallel 6.1.1 → 6.2.1 (Interactive Parallel Computing with IPython)

• ipython 6.2.1 → 6.4.0 (Enhanced Python shell)

• ipython_sql 0.3.8 → 0.3.9 (RDBMS access via IPython)

• ipywidgets 7.2.0 → 7.2.1 (IPython HTML widgets for Jupyter)

• jedi 0.11.1 → 0.12.0 (An autocompletion tool for Python that can be used for text editors)

• joblib 0.11 → 0.12.0 (Lightweight pipelining: using Python functions as pipeline jobs.)

• jupyterlab 0.31.12 → 0.32.1 (Jupyter lab environment notebook server extension)

• keras 2.1.5 → 2.2.0 (Theano-based Deep Learning library)

• lmfit 0.9.9 → 0.9.10 (Least-Squares Minimization with Bounds and Constraints)

• lxml 4.2.1 → 4.2.2 (Powerful and Pythonic XML processing library combining libxml2/libxslt with the ElementTree API.)

• monotonic 1.4 → 1.5 (An implementation of time.monotonic() for Python 2 & < 3.3)

• more_itertools 4.1.0 → 4.2.0 (More routines for operating on iterables, beyond itertools)

• moviepy 0.2.3.2 → 0.2.3.5 (Video editing with Python)

• netcdf4 1.3.1 → 1.4.0 (Provides an object-oriented python interface to the netCDF version 4 library)

• nltk 3.2.5 → 3.3 (The Natural Language Toolkit (NLTK) is a Python package for natural language processing.)

• notebook 5.4.1 → 5.5.0 (# Jupyter Notebook)

• numcodecs 0.5.4 → 0.5.5 (buffer compression and transformation codecs for use in data storage and communication applications)

• numexpr 2.6.4 → 2.6.5 (Fast evaluation of array expressions elementwise by using a vector-based virtual machine)

• numpy 1.14.2+mkl → 1.14.5+mkl (NumPy: multidimensional array processing for numbers, strings, records and objects (SciPy''s core module))

• numpydoc 0.7.0 → 0.8.0 (Sphinx extension to support docstrings in Numpy format)

• palettable 3.1.0 → 3.1.1 (Color palettes for Python)

• pandas 0.22.0 → 0.23.1 (Powerful data structures for data analysis, time series and statistics)

• param 1.5.2 → 1.6.1 (Declarative Python programming using Parameters.)

• parso 0.1.1 → 0.2.1 (A Python Parser)

• pexpect 4.4.0 → 4.6.0+dummy (Pexpect allows easy control of interactive console applications.)

• pillow 5.0.0 → 5.1.1 (Python Imaging Library (fork))

• pip 9.0.3 → 10.0.1 (A tool for installing and managing Python packages)

• psutil 5.4.3 → 5.4.6 (Provides an interface for retrieving information on all running processes and system utilization (CPU, disk, memory, network) in a portable way)

• pybind11 2.2.2 → 2.2.3 (Seamless operability between C++11 and Python)

• pycodestyle 2.3.1 → 2.4.0 (Python style guide checker)

• pyflakes 1.6.0 → 2.0.0 (passive checker of Python programs)

• pylint 1.8.3 → 1.9.2 (Logilab code analysis module: analyzes Python source code looking for bugs and signs of poor quality)

• pymc3 3.3 → 3.4.1 (Markov Chain Monte Carlo sampling toolkit.)

• pyqtgraph 0.10.1.dev1 → 0.10.1.dev2 (Scientific Graphics and GUI Library for Python)

• pytest 3.5.0 → 3.6.2 (pytest: simple powerful testing with Python)

• python_dateutil 2.7.2 → 2.7.3 (Powerful extensions to the standard datetime module)

• pytz 2018.3 → 2018.4 (World Timezone Definitions for Python)

• pywin32 223 → 223.1 (Python library for Windows)

• pywinpty 0.5.1 → 0.5.4 (Python bindings for the winpty library)

• qtawesome 0.4.4 → 0.5.0.dev0 (FontAwesome icons in PyQt and PySide applications)

• qtpy 1.4.0 → 1.4.2 (Provides an abstraction layer on top of the various Qt bindings (PyQt5, PyQt4 and PySide) and additional custom QWidgets.)

• regex 2018.2.21 → 2018.6.21 (Alternative regular expression module, to replace re.)

• requests 2.18.4 → 2.19.1 (Requests is an Apache2 Licensed HTTP library, written in Python, for human beings.)

• rope 0.10.5 → 0.10.7 (a python refactoring library...)

• rpy2 2.9.2 → 2.9.3 (Python interface to the R language (embedded R))

• scikit_image 0.13.1 → 0.14.0 (Image processing toolbox for SciPy)

• scipy 1.0.1 → 1.1.0 (SciPy: Scientific Library for Python (advanced math, signal processing, optimization, statistics, ...))

• setuptools 39.0.1 → 39.2.0 (Download, build, install, upgrade, and uninstall Python packages - easily)

• simplejson 3.13.2 → 3.14.0 (Simple, fast, extensible JSON (JavaScript Object Notation) encoder/decoder)

• sortedcontainers 1.5.9 → 2.0.4 (Python Sorted Container Types: SortedList, SortedDict, and SortedSet)

• sounddevice 0.3.10 → 0.3.11 (Play and Record Sound with Python)

• sphinx 1.7.2 → 1.7.5 (Tool for generating documentation which uses reStructuredText as its markup language)

• sphinx_rtd_theme 0.2.4 → 0.4.0 (ReadTheDocs.org theme for Sphinx, 2013 version.)

• sphinxcontrib_websupport 1.0.1 → 1.1.0 (Sphinx API for Web Apps)

• sqlalchemy 1.2.6 → 1.2.8 (SQL Toolkit and Object Relational Mapper)

• statsmodels 0.8.0 → 0.9.0 (Statistical computations and models for use with SciPy)

• tables 3.4.2 → 3.4.4 (Package based on HDF5 library for managing hierarchical datasets (extremely large amounts of data))

• theano 1.0.1 → 1.0.2 (Optimizing compiler for evaluating mathematical expressions on CPUs and GPUs.)

• tornado 4.5.3 → 5.0.2 (Scalable, non-blocking web server and tools (required for IPython notebook))

• tqdm 4.19.9 → 4.23.4 (A Simple Python Progress Meter)

• traittypes 0.0.6 → 0.2.1 (Scipy trait types)

• urllib3 1.21.1 → 1.23 (HTTP library with thread-safe connection pooling, file post, and more.)

• vega 0.5.0 → 1.3.0 (An IPython/ Jupyter widget for Vega and Vega-Lite)

• vega3 0.5.0 → 0.13.0 (An IPython/ Jupyter widget for Vega 3 and Vega-Lite 2)

• wheel 0.30.0 → 0.31.1 (A built-package format for Python.)

• widgetsnbextension 3.2.0 → 3.2.1 (IPython HTML widgets for Jupyter)

• xarray 0.10.2 → 0.10.7 (N-D labeled arrays and datasets in Python)

• xlsxwriter 1.0.2 → 1.0.5 (A Python module for creating Excel XLSX files.)

Removed packages:

• absl_py 0.1.11 (Abseil Python Common Libraries)

• astor 0.6.2 (Read/rewrite/write Python ASTs)

• gast 0.2.0 (Python AST that abstracts the underlying Python version)

• grpcio 1.10.0 (HTTP/2-based RPC framework)

• protobuf 3.5.1 (Protocol Buffers - Google's data interchange format)

• ptyprocess 0.5.2 (Run a subprocess in a pseudo terminal)

• tensorboard 1.7.0 (TensorBoard lets you watch Tensors Flow)

• tensorflow 1.7.0 (TensorFlow helps the tensors flow)

• termcolor 1.1.0 (ANSII Color formatting for output in terminal)

• vega_datasets 0.4.1 (A Python package for offline access to Vega datasets)

You should call the package PyTorch and not Torch.

the name of packages is the one you see per pip list.

@stonebig ,
In your package list you say the Numpy is Numpy + Intel MKL.

Yet if I do np.__config__.show() I get:

blas_mkl_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_blas95_lp64', 'mkl_rt']
    library_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64_win']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/include']
blas_opt_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_blas95_lp64', 'mkl_rt']
    library_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64_win']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/include']
lapack_mkl_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_blas95_lp64', 'mkl_rt']
    library_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64_win']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/include']
lapack_opt_info:
    libraries = ['mkl_lapack95_lp64', 'mkl_blas95_lp64', 'mkl_rt']
    library_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/lib/intel64_win']
    define_macros = [('SCIPY_MKL_H', None), ('HAVE_CBLAS', None)]
    include_dirs = ['C:/Program Files (x86)/IntelSWTools/compilers_and_libraries_2018/windows/mkl/include']

Which means Intel MKL isn't installed (It looks in the paths it might installed, it is not part of the package).
Could we have Numpy + MKL like in Anaconda?
It will improve performance greatly!

hum, wasn't it like that also in previous versions ?

I actually not sure what it means.
Could it be just a log of the configuration of the computer used to compile it?

I am not an expert on that.

I saw that you use Python Extension Packages for Windows - Christoph Gohlke.

Maybe it is better asking him (@cgohlke) about this (https://github.com/cgohlke)?

Spyder v3.3.0 is released: https://github.com/spyder-ide/spyder/releases

spyder-3.3.0 is there. trying to fix the last issue before rc bloomberg/bqplot#671

rc : 2018-07-08 (python-3.7.0, spyder-3.3.0)

Changes from b4:

• Python-3.7.0 final, Python-3.6.6, spyder-3.3.0, pandas-0.23.3
• holoviews-1.11.0a1, hvplot-0.2.0 (renamed from holoplot), ipyleaflet-0.9.0
• nota on Python-3.7.0:
  . a persistant glitch on Python-3.7.0 64bit/Pandas-0.23.x: ioam/holoviews#2738
  . pythonnet miss the python-3.7 mark (was waiting appveyor to provide a python-3.7.0 environment)

Areas of particular interest for testers:

• stable and running,

History of changes for WinPython-64bit 3.6.6.1Qt5rc

The following changes were made to WinPython-64bit distribution since version 3.6.5.1Qt5.

Python packages

New packages:

• appdirs 1.4.3 (A small Python module for determining appropriate " + "platform-specific dirs, e.g. a "user data dir".)

• hvplot 0.2.0 (A high-level plotting API for pandas, dask, streamz and xarray built on HoloViews)

• intake 0.1.3 (Data input plugin and catalog system)

• keras_applications 1.0.2 (Reference implementations of popular deep learning models)

• keras_preprocessing 1.0.1 (Easy data preprocessing and data augmentation for deep learning models)

• keyring 13.2.1 (Store and access your passwords safely.)

• msgpack_numpy 0.4.3 (Numpy data serialization using msgpack)

• msgpack_python 0.5.4+dummy (MessagePack (de)serializer.)

• parambokeh 0.2.2 (Declarative Python programming using Parameters.)

• pyviz_comms 0.1.1 (Launch jobs, organize the output, and dissect the results)

• pywin32_ctypes 0.1.2 (A (partial) reimplementation of pywin32 that is pure python (uses ctypes/cffi))

• ruamel.yaml 0.15.42 (a YAML parser/emitter that supports roundtrip preservation of comments, seq/map flow style, and map key order)

• spyder_kernels 0.2.4 (Jupyter kernels for the Spyder console)

Upgraded packages:

• alabaster 0.7.10 → 0.7.11 (A configurable sidebar-enabled Sphinx theme)

• altair 2.0.1 → 2.1.0 (High-level declarative visualization library for Python)

• astroid 1.6.4 → 1.6.5 (Rebuild a new abstract syntax tree from Python's ast (required for pylint))

• bloscpack 0.11.0 → 0.13.0 (Command line interface to and serialization format for Blosc)

• bokeh 0.12.16 → 0.13.0 (Statistical and novel interactive HTML plots for Python)

• cvxpy 1.0.3 → 1.0.6 (A domain-specific language for modeling convex optimization problems in Python)

• dask 0.17.5 → 0.18.1 (Minimal task scheduling abstraction)

• datashader 0.6.6 → 0.6.7 (Data visualization toolchain based on aggregating into a grid)

• dill 0.2.7.1 → 0.2.8.2 (serialize all of python (almost))

• distributed 1.21.8 → 1.22.0 (Distributed computing)

• holoviews 1.10.5a1 → 1.11.0a1 (Composable, declarative data structures for building complex visualizations easily.)

• idna 2.6 → 2.7 (Internationalized Domain Names in Applications (IDNA))

• ipyleaflet 0.8.1 → 0.9.0 (A Jupyter widget for dynamic Leaflet maps)

• ipympl 0.1.1 → 0.2.0 (Matplotlib Jupyter Extension)

• ipyparallel 6.1.1 → 6.2.2 (Interactive Parallel Computing with IPython)

• ipyscales 0.1.3 → 0.2.2 (A widget library for scales)

• jedi 0.12.0 → 0.12.1 (An autocompletion tool for Python that can be used for text editors)

• joblib 0.11 → 0.12.0 (Lightweight pipelining: using Python functions as pipeline jobs.)

• keras 2.1.6 → 2.2.0 (Theano-based Deep Learning library)

• llvmlite 0.23.2 → 0.24.0 (lightweight wrapper around basic LLVM functionality)

• lmfit 0.9.10 → 0.9.11 (Least-Squares Minimization with Bounds and Constraints)

• loky 2.1.1 → 2.1.4 (Robust and reusable Executor for joblib)

• lxml 4.2.1 → 4.2.3 (Powerful and Pythonic XML processing library combining libxml2/libxslt with the ElementTree API.)

• multiprocess 0.70.5 → 0.70.6.1 (better multiprocessing and multithreading in python)

• mypy 0.600 → 0.610 (Optional static typing for Python)

• numba 0.38.1 → 0.39.0 (compiling Python code using LLVM)

• numpy 1.14.3+mkl → 1.14.5+mkl (NumPy: multidimensional array processing for numbers, strings, records and objects (SciPy''s core module))

• octave_kernel 0.28.3 → 0.28.4 (A Jupyter kernel for Octave.)

• pandas 0.23.0 → 0.23.3 (Powerful data structures for data analysis, time series and statistics)

• param 1.6.1 → 1.7.0 (Declarative Python programming using Parameters.)

• parso 0.2.1 → 0.3.0 (A Python Parser)

• pillow 5.1.1 → 5.2.0 (Python Imaging Library (fork))

• psutil 5.4.5 → 5.4.6 (Provides an interface for retrieving information on all running processes and system utilization (CPU, disk, memory, network) in a portable way)

• py 1.5.3 → 1.5.4 (library with cross-python path, ini-parsing, io, code, log facilities)

• pyct 0.4.2 → 0.4.3 (python package common tasks for users (e.g. copy examples, fetch data, ...))

• pylint 1.9.1 → 1.9.2 (Logilab code analysis module: analyzes Python source code looking for bugs and signs of poor quality)

• pymongo 3.6.1 → 3.7.0 (Python driver for MongoDB http://www.mongodb.org)

• pyqtgraph 0.10.1.dev2 → 0.11.0.dev0 (Scientific Graphics and GUI Library for Python)

• pytest 3.6.0 → 3.6.3 (pytest: simple powerful testing with Python)

• Python 3.6.5 → 3.6.6 (Python programming language with standard library)

• python_snappy 0.5.2 → 0.5.3 (Python library for the snappy compression library from Google)

• pythreejs 1.0.0 → 1.1.0 (Interactive 3d graphics for the Jupyter notebook, using Three.js from Jupyter interactive widgets.)

• pytz 2018.4 → 2018.5 (World Timezone Definitions for Python)

• pywin32 223 → 223.1 (Python library for Windows)

• pywinpty 0.5.3 → 0.5.4 (Python bindings for the winpty library)

• pyyaml 3.12 → 4.1 (YAML parser and emitter for Python)

• regex 2018.2.21 → 2018.6.21 (Alternative regular expression module, to replace re.)

• requests 2.18.4 → 2.19.1 (Requests is an Apache2 Licensed HTTP library, written in Python, for human beings.)

• rpy2 2.9.3 → 2.9.4 (Python interface to the R language (embedded R))

• simplejson 3.14.0 → 3.16.0 (Simple, fast, extensible JSON (JavaScript Object Notation) encoder/decoder)

• sortedcontainers 2.0.2 → 2.0.4 (Python Sorted Container Types: SortedList, SortedDict, and SortedSet)

• sphinx_rtd_theme 0.3.1 → 0.4.0 (ReadTheDocs.org theme for Sphinx, 2013 version.)

• sphinxcontrib_websupport 1.0.1 → 1.1.0 (Sphinx API for Web Apps)

• spyder 3.2.8 → 3.3.0 (Scientific PYthon Development EnviRonment: designed for interactive computing and data visualisation with a simple and intuitive user interface)

• sqlalchemy 1.2.8 → 1.2.9 (SQL Toolkit and Object Relational Mapper)

• tables 3.4.3 → 3.4.4 (Package based on HDF5 library for managing hierarchical datasets (extremely large amounts of data))

• traittypes 0.2.0 → 0.2.1 (Scipy trait types)

• urllib3 1.21.1 → 1.23 (HTTP library with thread-safe connection pooling, file post, and more.)

• vega 1.1.1 → 1.3.0 (An IPython/ Jupyter widget for Vega and Vega-Lite)

• winpython 1.10.20180404 → 1.10.20180624 (WinPython distribution tools, including WPPM (package manager))

• xarray 0.10.4 → 0.10.7 (N-D labeled arrays and datasets in Python)

Removed packages:

• blaze 0.10.1 (Blaze)

• flask_cors 3.0.4 (A Flask extension adding a decorator for CORS support)

• odo 0.5.0 (Data migration in Python)

@stonebig , Any news on the MKL thing?

no. all my concentration is on understanding bloomberg/bqplot#671, now resolved.

WinPython-64bit 3.6.6.1Qt5rc is working well for me so far. I am experiencing some bugs in Spyder v3.3.0 that are planned to be fixed in v3.3.1. Since v3.3.1 may be 4 months from now, I might update Spyder to a development version before then. If this has problems, I want to be able restore Spyder v3.3.0.

@stonebig - Since Spyder v3.3.0 is not available at https://www.lfd.uci.edu/~gohlke/pythonlibs/ did you download wheels for Spyder and Spyder-kernels from PyPI?

for the rc, it was the PyPi standard wheels.