4ibd-s1-project-ml
| README.md <= Read it now
| .gitignore <= Don't forget using this
|
└─── default <= Folder of Project
| |
| └─── notebooks <= Put your notebook on this folder
| | | largerCNN.ipynb <= Test of Larger Convolutional Neuronal Network
| | | SampleCNN.ipynb <= Test of Sample Convolutional Neuronal Network
| | | MLP.ipynb <= Test of Multi Layer Perceptron
| | | SLP.ipynb <= Test of Sample perceptron
| | |
| | └─── tensorboardNotebooks <= Tensorboard Folder generate by your notebook
| | | └─── logs <= Logs of model Tensorboard
| | |
| | └─── testModel <= You can create many notebook in this folder to Test model
| | └─── Test <= Stock your test of this folder
| | (this folder will be delete when we doing last commit of project)
| |
| └─── src <= Folder of model(mlp,slp,cnn,rnn) type .py
| | └─── mModel <= Class of Run Many Models Machine/Deep learning
| | main.py
| | README.md
| |
| └─── mModel <= Class of Model
| └─── model <= Folder of Model
| | modelManager.py <= Class to manage data, preprocessing, tensorboard...
| | cnn.py <= Convolutional Neuronal Network Class
| | mlp.py <= Multi Layer Perceptron Class
| | slp.py <= Sample Layer Class
| | rnn.py <= Recurrent Neural Networks Class
|
└─── utils
| | README.md <= Readme about Docker
| |
| └─── anaconda3 <= Anaconda3 Docker
| | | docker-compose.yml
| | | docker.env
| | |
| | └─── jupyter <= Folder of jupyter
| | | Dockerfile
| |
| └─── python3 <= Python3 Docker
| | docker.env <= Environnement variables
| | docker-compose.yml <= Configuration docker
| |
| └─── python
| | Dockerfile <= Dockerfile settings
| | requirements.txt <= List of libraries to your python
python -m default.apps.src.mModel.main "model_name" nbr_epochs
docker exec container_name python -m default.apps.src.mModel.main "model_name" epochs
Open a terminal window in your root project directory. Run:
tensorboard --logdir path
Go to the URL it provides OR on windows:
http://localhost:6006/
from 4ibd-s1-project-ml run :
mlflow ui
mlflow ui web :
http://localhost:5000/
After having installed and configured Keras on their machine, students will have to study different models on Datasets by varying the different hyper parameters of these datasets:
-
Previous models
- Linear Model
- Perceptron Multilayer
-
New models
- Conv Net (s)
- ResNets / HighwayNets - RNN (s)
Students will be strongly advised to rely on Tensorboard to visualize compare and retranscribe the performances of these models. It will be important to show for each dataset, for each model: The influence of all the hyperparameters of the models (structure, activation functions, ...), as well as the parameters of the learning algorithms (learning rate, momentum, ...)
Reference books (books, articles, magazines, websites ...)
Computer tools to install Keras / Tensorflow / Jupyter
link SDK to Google Cloud Plateform : global link : https://cloud.google.com/sdk/?hl=fr
ssh-keygen -t rsa
After you confirm the passphrase, the system generates the key pair.
Your identification has been saved in /Users/user/.ssh/id_rsa.
Your public key has been saved in /Users/user/.ssh/id_rsa.pub.
The key fingerprint is:
ae:89:72:0b:85:da:5a:f4:7c:1f:c2:43:fd:c6:44:38 [email protected]
The key's randomart image is:
+--[ RSA 2048]----+
| |
| . |
| E . |
| . . o |
| o . . S . |
| + + o . + |
|. + o = o + |
| o...o * o |
|. oo.o . |
+-----------------+
pbcopy < ~/.ssh/id_rsa.pub
ssh-add ~/.ssh/id_rsa
ssh-add -L
tutorial link
You need create your google account (@gmail) and go to the link and click in "console"
after that you must create your Google Cloud Plateform account. Now you need create your VM instance, so your need go to navigator menu (left of browser), and go in compute engine->VM instance.
Step to Create VM :
- Clic on create VM
- clic on marketplace
- choose "machinelearning" and "free"
- clic on "AISE TensorFlow NVidia GPU Production"
- clic on "lautch one compute engine"
- you can change name of instance and add ram
- choose "nvidia-tesla-p100" GPU
- clic on "deploy"
