BigData_training
Big Data Essentials: HDFS, MapReduce and Spark RDD by Yandex
1.Hadoop Yarn Notebook
Docker container with Hadoop Yarn Jupyter Notebook: yarn-notebook
To SSH into the container: $docker exec -it hadoop_8881 bash
Spark into Jupyter Notebook (PySpark)
Docker container with Spark Jupyter Notebook: spark-course1
docker run -it -p 8880:8888 --name spark_8880 bigdatateam/spark-course1
To SSH into the container: $ docker exec -it spark_8880 bash
Big Data Applications: Machine Learning at Scale by Yandex
2.Docker container Spark course 3: spark-course1
Introduction to Deep Learning por National Research University Higher School of Economics
3.https://github.com/hse-aml/intro-to-dl
intro-to-dl
_________________________________ at folder Docker container with Jupyter Environment: coursera-aml-docker
with source repository https://github.com/ZEMUSHKA/coursera-aml-docker
Week 3:
- 3.1 Introduction to CNN
- 3.2 Modern CNNs
Your first CNN on CIFAR-10
- Follow the instructions on https://hub.docker.com/r/zimovnov/coursera-aml-docker/ to install Docker container with all necessary software installed. After that you should see a Jupyter page in your browser.
docker run -it -p 8882:8080 --name coursera-aml-1 zimovnov/coursera-aml-docker
Add -p 7007:7007
if wanting to access Tensorflow dashboard.
- Or buid it from the Dockerfile at folder
coursera-aml-docker
:
docker build -t brjapon/coursera-aml-docker .
docker run -it -p 8882:8080 --name coursera-aml-10 brjapon/coursera-aml-docker
- SSH into the container and clone the repo with the exercises:
docker exec -it coursera-aml-1 bash
git clone https://github.com/hse-aml/intro-to-dl
-
Download Keras and week 3 resources, by executing the required lines in notebook
./intro-to-dl/download_resources.ipynb
-
Run notebook Task 1
./intro-to-dl/week3/week3_task1_first_cnn_cifar10_clean.ipynb
-
Run notebook Task 2
./intro-to-dl/week3/week3_task2_fine_tuning_clean.ipynb
Container checkpoints
You might want to make a checkpoint of your work so that you can return to it later. Think of it as a backup or commit in version control system.
Saving container state
You will first have to stop the container following instructions above. Now you need to save the container state so that you can return to it later
docker commit coursera-aml-1 coursera-aml-snap-1
You can make sure that it's saved by running docker images.
Creating new container from previous checkpoint If you want to continue working from a particular checkpoint, you should run a new container from your saved image by executing
docker run -it -p 8882:8080 -p 7007:7007 --name coursera-aml-2 coursera-aml-snap-1
Notice that we incremented index in the container name, because we created a new container.
Using GPU in your container
You can use NVIDIA GPU in your container on Linux host machine.
Setup docker following instructions from NVIDIA:
https://github.com/NVIDIA/nvidia-docker/wiki/Installation-(version-2.0)#prerequisites In your container replace CPU TensorFlow version with the one that supports GPU:
pip3 uninstall tensorflow
pip3 install tensorflow-gpu==1.2.1
You will also have to install NVIDIA GPU driver, CUDA toolkit and CuDNN (requires registration with NVIDIA) in your container in order for TensorFlow to work with your GPU: https://www.tensorflow.org/versions/r1.2/install/install_linux#nvidia_requirements_to_run_tensorflow_with_gpu_support
It can be hard to follow, so you might choose to stick to a CPU version, which is also fine for the purpose of this course. TensorFlow provides Docker files with TensorFlow on GPU, but they don't have all the additional dependencies we need, this is for advanced users: https://github.com/tensorflow/tensorflow/tree/master/tensorflow/tools/docker
- 3.3 Application of CNNs
Applied AI with DeepLearning by IBM
4.https://github.com/romeokienzler/developerWorks