Prepare for article release

README.md

@@ -37,3 +37,14 @@ floydhub/PROJECT_NAME:VERSION-ENV-gpu
 ```
 
 Contains docker images for popular deep learning frameworks including: Tensorflow, PyTorch and Torch.
+
+## Repository's structure
+
+Add image from the blog & link
+
+TODO: Floydker link
+
+TODO: CI link?
+
+TODO: base & dl explanation (copy from the article)
+

base/buildpack-deps/Dockerfile

@@ -1,4 +1,5 @@
 FROM buildpack-deps:bionic
+MAINTAINER FloydHub "[email protected]"
 
 RUN apt-get update && apt-get install -y --no-install-recommends \
 		sudo \

dl/dl-base/dl-base-2.x.x.jinja

@@ -64,7 +64,9 @@ RUN pip --no-cache-dir install \
 
 
 # Install and Enable jupyter-widgets - see https://github.com/jupyter-widgets/ipywidgets/tree/master/packages/jupyterlab-manager
-RUN jupyter labextension install @jupyter-widgets/[email protected]
+RUN jupyter labextension install \
+        @jupyter-widgets/[email protected] \
+        jupyterlab-plotly
 
 {%- if arch == 'gpu' %}
 # Update Cmake to the latest version to build xgboost on CUDA

dl/tensorflow/tests/2.0/run.sh

@@ -1,9 +1,9 @@
 #! /bin/bash
 #############################
-#   TF 1.7 Testing Script   #
+#   TF 2.0 Testing Script   #
 #############################
 
-#TF Session
+# TF Session
 python tf_keras_mnist.py
 
 # TF Eager
@@ -13,56 +13,4 @@ python mnist_eager.py
 python keras_mnist_test.py
 
 
-#TODO Move from tf.tf_cnn_benchmarks to perfzero https://github.com/tensorflow/benchmarks/tree/master/perfzero
-
-# # Delete benchmarks folder if exists
-# if test -d benchmarks; then
-# 	rm -rf benchmarks
-# fi
-
-# Clone Benchmarks and go to a certain commit
-# git clone https://github.com/tensorflow/benchmarks.git
-# cd benchmarks/scripts/tf_cnn_benchmarks
-
-# # Check if GPU is installed (quick hack)
-# which nvidia-smi &> /dev/null
-# ISGPU=$?
-
-# if test $ISGPU -eq 1; then
-# 	# CPU or CPU2?
-# 	NCORE=`cat /proc/cpuinfo | grep processor | wc -l`
-
-# 	# Quick Benchmark on AlexNet (syntethic images, no data transformation, channel last)
-# 	if test $NCORE -eq 2; then
-# 		# CPU test (defined initial lr otherwise Loss is NaN )
-# 		python tf_cnn_benchmarks.py --device=cpu  --kmp_blocktime=0 --nodistortions --model=alexnet --data_format=NHWC --batch_size=32  --num_inter_threads=1 --num_intra_threads=$NCORE --init_learning_rate=0.00001
-# 	else
-# 		# CPU2 test
-# 		python tf_cnn_benchmarks.py --device=cpu  --kmp_blocktime=0 --nodistortions --model=alexnet --data_format=NHWC --batch_size=64  --num_inter_threads=1 --num_intra_threads=$NCORE --init_learning_rate=0.00001
-# 	fi
-# else
-# 	GPU_DEV_NAME=$(python -c 'from __future__ import print_function; import tensorflow as tf; print(tf.test.gpu_device_name())' 2> /dev/null)
-# 	case ${GPU_DEV_NAME} in
-# 	/device:GPU:*)
-# 		echo "GPU load test passed. Device found: ${GPU_DEV_NAME}"
-# 		;;
-# 	*)
-# 		echo "ERROR: Tensorflow not able to find GPU device! Device found: ${GPU_DEV_NAME}"
-# 		exit 1
-# 		;;
-# 	esac
-
-# 	# GPU or GPU2?
-# 	nvidia-smi | grep -q V100
-# 	ISV100=$?
-
-# 	# Quick Benchmark on Resnet-50 (syntethic images, no data transformation, channel first)
-# 	if test $ISV100 -eq 1; then
-# 		# GPU test
-# 		python tf_cnn_benchmarks.py --device=gpu --num_gpus=1 --batch_size=64 --model=resnet50 --nodistortions --data_format=NCHW
-# 	else
-# 		# GPU2 test (with MXP)
-# 		python tf_cnn_benchmarks.py --device=gpu --num_gpus=1 --batch_size=256 --model=resnet50 --nodistortions --data_format=NCHW --use_fp16=true
-# 	fi
-# fi
-
+# Consider to move from tf.tf_cnn_benchmarks to perfzero https://github.com/tensorflow/benchmarks/tree/master/perfzero