Check notebook after ci update

b0e3ba38 · Jean-Luc Parouty · 7c2ca0c7 · b0e3ba38 · b0e3ba38 · b0e3ba38
Commit b0e3ba38 authored 4 years ago by Jean-Luc Parouty
--- a/AE/01-AE-with-MNIST==done==.ipynb
+++ b/AE/01-AE-with-MNIST==done==.ipynb
--- a/AE/02-AE-with-MNIST-post==done==.ipynb
+++ b/AE/02-AE-with-MNIST-post==done==.ipynb
--- a/BHPD/01-DNN-Regression==done==.ipynb
+++ b/BHPD/01-DNN-Regression==done==.ipynb
--- a/BHPD/02-DNN-Regression-Premium==done==.ipynb
+++ b/BHPD/02-DNN-Regression-Premium==done==.ipynb
--- a/GTSRB/01-Preparation-of-data.ipynb
+++ b/GTSRB/01-Preparation-of-data.ipynb
@@ -159,7 +159,7 @@
    "`scale` : 1 mean 100% of the dataset - set 0.1 for tests  \n",
    "`output_dir` : where to write enhanced dataset, could be :\n",
    " - `./data`, for tests purpose\n",
-    " - `<datasets_dir>/celeba/enhanced` to add clusters in your datasets dir.  \n",
+    " - `<datasets_dir>/GTSRB/enhanced` to add clusters in your datasets dir.  \n",
    " \n",
    "Uncomment the right lines according to what you want :"
   ]
@@ -1437,7 +1437,7 @@
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
-   "version": "3.7.9"
+   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 %% Cell type:markdown id: tags:
  
 <img width="800px" src="../fidle/img/00-Fidle-header-01.svg"></img>
  
 # <!-- TITLE --> [GTSRB1] - Dataset analysis and preparation
 <!-- DESC --> Episode 1 : Analysis of the GTSRB dataset and creation of an enhanced dataset
 <!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->
  
 ## Objectives :
 - Understand the **complexity associated with data**, even when it is only images
 - Learn how to build up a simple and **usable image dataset**
  
 The German Traffic Sign Recognition Benchmark (GTSRB) is a dataset with more than 50,000 photos of road signs from about 40 classes.
 The final aim is to recognise them !
  
 Description is available there : http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset
  
  
 ## What we're going to do :
  
 - Understanding the dataset
 - Preparing and formatting enhanced data
 - Save enhanced datasets in h5 file format
  
 %% Cell type:markdown id: tags:
  
 ## Step 1 -  Import and init
  
 %% Cell type:code id: tags:
  
 ``` python
 import os, time, sys
 import csv
 import math, random
  
 import numpy as np
 import pandas as pd
 import matplotlib.pyplot as plt
 import h5py
  
 from skimage.morphology import disk
 from skimage.util import img_as_ubyte
 from skimage.filters import rank
 from skimage import io, color, exposure, transform
  
 from importlib import reload
  
 sys.path.append('..')
 import fidle.pwk as pwk
  
 datasets_dir = pwk.init('GTSRB1')
 ```
  
 %% Output
  
  
    <br>**FIDLE 2020 - Practical Work Module**
  
    Version              : 1.2b1 DEV
    Notebook id          : GTSRB1
    Run time             : Monday 11 January 2021, 21:34:27
    TensorFlow version   : 2.2.0
    Keras version        : 2.3.0-tf
    Datasets dir         : /home/pjluc/datasets/fidle
    Run dir              : ./run
    Update keras cache   : False
  
 %% Cell type:markdown id: tags:
  
 ## Step 2 - Parameters
 The generation of datasets may require some time and space : **10' and 10 GB**.
  
 You can choose to perform tests or generate the whole enhanced dataset by setting the following parameters:
 `scale` : 1 mean 100% of the dataset - set 0.1 for tests
 `output_dir` : where to write enhanced dataset, could be :
 - `./data`, for tests purpose
- - `<datasets_dir>/celeba/enhanced` to add clusters in your datasets dir.
+ - `<datasets_dir>/GTSRB/enhanced` to add clusters in your datasets dir.
  
 Uncomment the right lines according to what you want :
  
 %% Cell type:code id: tags:
  
 ``` python
 # ---- For smart tests :
 #
 scale      = 0.1
 output_dir = './data'
  
 # ---- For a Full dataset generation :
 #
 # scale      = 1
 # output_dir = f'{datasets_dir}/GTSRB/enhanced'
 ```
  
 %% Cell type:markdown id: tags:
  
 Override parameters (batch mode) - Just forget this cell
  
 %% Cell type:code id: tags:
  
 ``` python
 pwk.override('scale', 'output_dir')
 ```
  
 %% Cell type:markdown id: tags:
  
 ## Step 3 - Read the dataset
 Description is available there : http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset
 - Each directory contains one CSV file with annotations : `GT-<ClassID>.csv` and the training images
 - First line is fieldnames: `Filename ; Width ; Height ; Roi.X1 ; Roi.Y1 ; Roi.X2 ; Roi.Y2 ; ClassId`
  
 ### 3.1 - Understanding the dataset
 The original dataset is in : **\<dataset_dir\>/GTSRB/origine.**
 There is 3 subsets : **Train**, **Test** and **Meta.**
 Each subset have an **csv file** and a **subdir** with **images**.
  
  
 %% Cell type:code id: tags:
  
 ``` python
 df = pd.read_csv(f'{datasets_dir}/GTSRB/origine/Test.csv', header=0)
 display(df.head(10))
 ```
  
 %% Output
  
  
 %% Cell type:markdown id: tags:
  
 ### 3.2 - Usefull functions
 A nice function to read a subset :
  
 %% Cell type:code id: tags:
  
 ``` python
 def  read_csv_dataset(csv_file):
    '''
    Reads traffic sign data from German Traffic Sign Recognition Benchmark dataset.
    Arguments:
        csv filename :  Description file, Example /data/GTSRB/Train.csv
    Returns:
        x,y          :  np array of images, np array of corresponding labels
    '''
  
    path = os.path.dirname(csv_file)
    name = os.path.basename(csv_file)
  
    # ---- Read csv file
    #
    df = pd.read_csv(csv_file, header=0)
  
    # ---- Get filenames and ClassIds
    #
    filenames = df['Path'].to_list()
    y         = df['ClassId'].to_list()
    x         = []
  
    # ---- Read images
    #
    for filename in filenames:
        image=io.imread(f'{path}/{filename}')
        x.append(image)
        pwk.update_progress(name,len(x),len(filenames))
  
    # ---- Return
    #
    return np.array(x,dtype=object),np.array(y)
 ```
  
 %% Cell type:markdown id: tags:
  
 ### 3.2 - Read the data
 We will read the following datasets:
 - **Train** subset, for learning data as :  `x_train, y_train`
 - **Test** subset, for validation data as :  `x_test, y_test`
 - **Meta** subset, for visualisation as : `x_meta, y_meta`
  
 The learning data will be randomly mixted and the illustration data (Meta) sorted.
 Will take about 1'30s on HPC.
  
 %% Cell type:code id: tags:
  
 ``` python
 pwk.chrono_start()
  
 # ---- Read datasets
  
 (x_train,y_train) = read_csv_dataset(f'{datasets_dir}/GTSRB/origine/Train.csv')
 (x_test ,y_test)  = read_csv_dataset(f'{datasets_dir}/GTSRB/origine/Test.csv')
 (x_meta ,y_meta)  = read_csv_dataset(f'{datasets_dir}/GTSRB/origine/Meta.csv')
  
 # ---- Shuffle train set
  
 x_train, y_train = pwk.shuffle_np_dataset(x_train, y_train)
  
 # ---- Sort Meta
  
 combined = list(zip(x_meta,y_meta))
 combined.sort(key=lambda x: x[1])
 x_meta,y_meta = zip(*combined)
  
 pwk.chrono_show()
 ```
  
 %% Output
  
    Train.csv        [########################################] 100.0% of 39209
    Test.csv         [########################################] 100.0% of 12630
    Meta.csv         [########################################] 100.0% of 43
    
    Duration :  00:00:31 663ms
  
 %% Cell type:markdown id: tags:
  
 ## Step 4 - Few statistics about train dataset
 We want to know if our images are homogeneous in terms of size, ratio, width or height.
  
 ### 4.1 - Do statistics
  
 %% Cell type:code id: tags:
  
 ``` python
 train_size  = []
 train_ratio = []
 train_lx    = []
 train_ly    = []
  
 test_size   = []
 test_ratio  = []
 test_lx     = []
 test_ly     = []
  
 for image in x_train:
    (lx,ly,lz) = image.shape
    train_size.append(lx*ly/1024)
    train_ratio.append(lx/ly)
    train_lx.append(lx)
    train_ly.append(ly)
  
 for image in x_test:
    (lx,ly,lz) = image.shape
    test_size.append(lx*ly/1024)
    test_ratio.append(lx/ly)
    test_lx.append(lx)
    test_ly.append(ly)
 ```
  
 %% Cell type:markdown id: tags:
  
 ### 4.2 - Show statistics
  
 %% Cell type:code id: tags:
  
 ``` python
 # ------ Global stuff
 print("x_train shape : ",x_train.shape)
 print("y_train shape : ",y_train.shape)
 print("x_test  shape : ",x_test.shape)
 print("y_test  shape : ",y_test.shape)
  
 # ------ Statistics / sizes
 plt.figure(figsize=(16,6))
 plt.hist([train_size,test_size], bins=100)
 plt.gca().set(title='Sizes in Kpixels - Train=[{:5.2f}, {:5.2f}]'.format(min(train_size),max(train_size)),
              ylabel='Population', xlim=[0,30])
 plt.legend(['Train','Test'])
 pwk.save_fig('01-stats-sizes')
 plt.show()
  
 # ------ Statistics / ratio lx/ly
 plt.figure(figsize=(16,6))
 plt.hist([train_ratio,test_ratio], bins=100)
 plt.gca().set(title='Ratio lx/ly - Train=[{:5.2f}, {:5.2f}]'.format(min(train_ratio),max(train_ratio)),
              ylabel='Population', xlim=[0.8,1.2])
 plt.legend(['Train','Test'])
 pwk.save_fig('02-stats-ratios')
 plt.show()
  
 # ------ Statistics / lx
 plt.figure(figsize=(16,6))
 plt.hist([train_lx,test_lx], bins=100)
 plt.gca().set(title='Images lx - Train=[{:5.2f}, {:5.2f}]'.format(min(train_lx),max(train_lx)),
              ylabel='Population', xlim=[20,150])
 plt.legend(['Train','Test'])
 pwk.save_fig('03-stats-lx')
 plt.show()
  
 # ------ Statistics / ly
 plt.figure(figsize=(16,6))
 plt.hist([train_ly,test_ly], bins=100)
 plt.gca().set(title='Images ly - Train=[{:5.2f}, {:5.2f}]'.format(min(train_ly),max(train_ly)),
              ylabel='Population', xlim=[20,150])
 plt.legend(['Train','Test'])
 pwk.save_fig('04-stats-ly')
 plt.show()
  
 # ------ Statistics / classId
 plt.figure(figsize=(16,6))
 plt.hist([y_train,y_test], bins=43)
 plt.gca().set(title='ClassesId', ylabel='Population', xlim=[0,43])
 plt.legend(['Train','Test'])
 pwk.save_fig('05-stats-classes')
 plt.show()
 ```
  
 %% Output
  
    x_train shape :  (39209,)
    y_train shape :  (39209,)
    x_test  shape :  (12630,)
    y_test  shape :  (12630,)
  

  

  

  

  

  
 %% Cell type:markdown id: tags:
  
 ## Step 5 - List of classes
 What are the 43 classes of our images...
  
 %% Cell type:code id: tags:
  
 ``` python
 pwk.plot_images(x_meta,y_meta, range(43), columns=8, x_size=2, y_size=2,
                                colorbar=False, y_pred=None, cm='binary', save_as='06-meta-signs')
 ```
  
 %% Output
  

  
 %% Cell type:markdown id: tags:
  
 ## Step 6 - What does it really look like
  
 %% Cell type:code id: tags:
  
 ``` python
 # ---- Get and show few images
  
 samples = [ random.randint(0,len(x_train)-1) for i in range(32)]
 pwk.plot_images(x_train,y_train, samples, columns=8, x_size=2, y_size=2,
                colorbar=False, y_pred=None, cm='binary', save_as='07-real-signs')
 ```
  
 %% Output
  

  
 %% Cell type:markdown id: tags:
  
 ## Step 7 - dataset cooking...
  
 Images **must** :
 - have the **same size** to match the size of the network,
 - be **normalized**.
  
 It is possible to work on **rgb** or **monochrome** images and to **equalize** the histograms.
  
 See : [Exposure with scikit-image](https://scikit-image.org/docs/dev/api/skimage.exposure.html)
 See : [Local histogram equalization](https://scikit-image.org/docs/dev/api/skimage.filters.rank.html#skimage.filters.rank.equalize)
 See : [Histogram equalization](https://scikit-image.org/docs/dev/api/skimage.exposure.html#skimage.exposure.equalize_hist)
  
 ### 7.1 - Enhancement cooking
  
 %% Cell type:code id: tags:
  
 ``` python
 def images_enhancement(images, width=25, height=25, mode='RGB'):
    '''
    Resize and convert images - doesn't change originals.
    input images must be RGBA or RGB.
    Note : all outputs are fixed size numpy array of float64
    args:
        images :         images list
        width,height :   new images size (25,25)
        mode :           RGB | RGB-HE | L | L-HE | L-LHE | L-CLAHE
    return:
        numpy array of enhanced images
    '''
    modes = { 'RGB':3, 'RGB-HE':3, 'L':1, 'L-HE':1, 'L-LHE':1, 'L-CLAHE':1}
    lz=modes[mode]
  
    out=[]
    for img in images:
  
        # ---- if RGBA, convert to RGB
        if img.shape[2]==4:
            img=color.rgba2rgb(img)
  
        # ---- Resize
        img = transform.resize(img, (width,height))
  
        # ---- RGB / Histogram Equalization
        if mode=='RGB-HE':
            hsv = color.rgb2hsv(img.reshape(width,height,3))
            hsv[:, :, 2] = exposure.equalize_hist(hsv[:, :, 2])
            img = color.hsv2rgb(hsv)
  
        # ---- Grayscale
        if mode=='L':
            img=color.rgb2gray(img)
  
        # ---- Grayscale / Histogram Equalization
        if mode=='L-HE':
            img=color.rgb2gray(img)
            img=exposure.equalize_hist(img)
  
        # ---- Grayscale / Local Histogram Equalization
        if mode=='L-LHE':
            img=color.rgb2gray(img)
            img = img_as_ubyte(img)
            img=rank.equalize(img, disk(10))/255.
  
        # ---- Grayscale / Contrast Limited Adaptive Histogram Equalization (CLAHE)
        if mode=='L-CLAHE':
            img=color.rgb2gray(img)
            img=exposure.equalize_adapthist(img)
  
        # ---- Add image in list of list
        out.append(img)
        pwk.update_progress('Enhancement: ',len(out),len(images))
  
    # ---- Reshape images
    #     (-1, width,height,1) for L
    #     (-1, width,height,3) for RGB
    #
    out = np.array(out,dtype='float64')
    out = out.reshape(-1,width,height,lz)
    return out
 ```
  
 %% Cell type:markdown id: tags:
  
 ### 7.2 - To get an idea of the different recipes
  
 %% Cell type:code id: tags:
  
 ``` python
 i=random.randint(0,len(x_train)-16)
 x_samples = x_train[i:i+16]
 y_samples = y_train[i:i+16]
  
 datasets  = {}
  
 datasets['RGB']      = images_enhancement( x_samples, width=25, height=25, mode='RGB'  )
 datasets['RGB-HE']   = images_enhancement( x_samples, width=25, height=25, mode='RGB-HE'  )
 datasets['L']        = images_enhancement( x_samples, width=25, height=25, mode='L'  )
 datasets['L-HE']     = images_enhancement( x_samples, width=25, height=25, mode='L-HE'  )
 datasets['L-LHE']    = images_enhancement( x_samples, width=25, height=25, mode='L-LHE'  )
 datasets['L-CLAHE']  = images_enhancement( x_samples, width=25, height=25, mode='L-CLAHE'  )
  
 pwk.subtitle('EXPECTED')
 x_expected=[ x_meta[i] for i in y_samples]
 pwk.plot_images(x_expected, y_samples, range(12), columns=12, x_size=1, y_size=1,
                colorbar=False, y_pred=None, cm='binary', save_as='08-expected')
  
 pwk.subtitle('ORIGINAL')
 pwk.plot_images(x_samples,  y_samples, range(12), columns=12, x_size=1, y_size=1,
                colorbar=False, y_pred=None, cm='binary', save_as='09-original')
  
 pwk.subtitle('ENHANCED')
 n=10
 for k,d in datasets.items():
    print("dataset : {}  min,max=[{:.3f},{:.3f}]  shape={}".format(k,d.min(),d.max(), d.shape))
    pwk.plot_images(d, y_samples, range(12), columns=12, x_size=1, y_size=1,
                    colorbar=False, y_pred=None, cm='binary', save_as=f'{n}-enhanced-{k}')
    n+=1
 ```
  
 %% Output
  
    Enhancement:     [################] 100.0% of 16
    Enhancement:     [################] 100.0% of 16
    Enhancement:     [################] 100.0% of 16
    Enhancement:     [################] 100.0% of 16
    Enhancement:     [################] 100.0% of 16
    Enhancement:     [################] 100.0% of 16
  
    <br>**EXPECTED**
  

  
    <br>**ORIGINAL**
  

  
    <br>**ENHANCED**
  
    dataset : RGB  min,max=[0.034,1.000]  shape=(16, 25, 25, 3)
  

  
    dataset : RGB-HE  min,max=[0.001,1.000]  shape=(16, 25, 25, 3)
  

  
    dataset : L  min,max=[0.042,1.000]  shape=(16, 25, 25, 1)
  

  
    dataset : L-HE  min,max=[0.002,1.000]  shape=(16, 25, 25, 1)
  

  
    dataset : L-LHE  min,max=[0.000,1.000]  shape=(16, 25, 25, 1)
  

  
    dataset : L-CLAHE  min,max=[0.000,1.000]  shape=(16, 25, 25, 1)
  

  
 %% Cell type:markdown id: tags:
  
 ### 7.3 - Cook and save
 A function to save a dataset
  
 %% Cell type:code id: tags:
  
 ``` python
 def save_h5_dataset(x_train, y_train, x_test, y_test, x_meta,y_meta, filename):
  
    # ---- Create h5 file
    with h5py.File(filename, "w") as f:
        f.create_dataset("x_train", data=x_train)
        f.create_dataset("y_train", data=y_train)
        f.create_dataset("x_test",  data=x_test)
        f.create_dataset("y_test",  data=y_test)
        f.create_dataset("x_meta",  data=x_meta)
        f.create_dataset("y_meta",  data=y_meta)
  
    # ---- done
    size=os.path.getsize(filename)/(1024*1024)
    print('Dataset : {:24s}  shape : {:22s} size : {:6.1f} Mo   (saved)'.format(filename, str(x_train.shape),size))
 ```
  
 %% Cell type:markdown id: tags:
  
 Generate enhanced datasets :
  
 %% Cell type:code id: tags:
  
 ``` python
 pwk.chrono_start()
  
 n_train = int( len(x_train)*scale )
 n_test  = int( len(x_test)*scale )
  
 pwk.subtitle('Parameters :')
 print(f'Scale is : {scale}')
 print(f'x_train length is : {n_train}')
 print(f'x_test  length is : {n_test}')
 print(f'output dir is     : {output_dir}\n')
  
 pwk.subtitle('Running...')
  
 pwk.mkdir(output_dir)
  
 for s in [24, 48]:
    for m in ['RGB', 'RGB-HE', 'L', 'L-LHE']:
        # ---- A nice dataset name
        filename = f'{output_dir}/set-{s}x{s}-{m}.h5'
        pwk.subtitle(f'Dataset : {filename}')
  
        # ---- Enhancement
        #      Note : x_train is a numpy array of python objects (images with <> sizes)
        #             but images_enhancement() return a real array of float64 numpy (images with same size)
        #             so, we can save it in nice h5 files
        #
        x_train_new = images_enhancement( x_train[:n_train], width=s, height=s, mode=m )
        x_test_new  = images_enhancement( x_test[:n_test],  width=s, height=s, mode=m )
        x_meta_new  = images_enhancement( x_meta,  width=s, height=s, mode='RGB' )
  
        # ---- Save
        save_h5_dataset( x_train_new, y_train[:n_train], x_test_new, y_test[:n_test], x_meta_new,y_meta, filename)
  
 x_train_new,x_test_new=0,0
  
 pwk.chrono_show()
 ```
  
 %% Output
  
    <br>**Parameters :**
  
    Scale is : 0.1
    x_train length is : 3920
    x_test  length is : 1263
    output dir is     : ./data
    
  
    <br>**Running...**
  
    <br>**Dataset : ./data/set-24x24-RGB.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-24x24-RGB.h5   shape : (3920, 24, 24, 3)      size :   68.9 Mo   (saved)
  
    <br>**Dataset : ./data/set-24x24-RGB-HE.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-24x24-RGB-HE.h5  shape : (3920, 24, 24, 3)      size :   68.9 Mo   (saved)
  
    <br>**Dataset : ./data/set-24x24-L.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-24x24-L.h5     shape : (3920, 24, 24, 1)      size :   23.4 Mo   (saved)
  
    <br>**Dataset : ./data/set-24x24-L-LHE.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-24x24-L-LHE.h5  shape : (3920, 24, 24, 1)      size :   23.4 Mo   (saved)
  
    <br>**Dataset : ./data/set-48x48-RGB.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-48x48-RGB.h5   shape : (3920, 48, 48, 3)      size :  275.6 Mo   (saved)
  
    <br>**Dataset : ./data/set-48x48-RGB-HE.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-48x48-RGB-HE.h5  shape : (3920, 48, 48, 3)      size :  275.6 Mo   (saved)
  
    <br>**Dataset : ./data/set-48x48-L.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-48x48-L.h5     shape : (3920, 48, 48, 1)      size :   93.4 Mo   (saved)
  
    <br>**Dataset : ./data/set-48x48-L-LHE.h5**
  
    Enhancement:     [########################################] 100.0% of 3920
    Enhancement:     [########################################] 100.0% of 1263
    Enhancement:     [########################################] 100.0% of 43
    Dataset : ./data/set-48x48-L-LHE.h5  shape : (3920, 48, 48, 1)      size :   93.4 Mo   (saved)
    
    Duration :  00:00:59 691ms
  
 %% Cell type:markdown id: tags:
  
 <div class='todo'>
  Adapt the code below to read :
  <ul>
      <li>the different h5 datasets you saved in ./data,</li>
      <li>The h5 datasets available in the Fidle project datasets directory.</li>
  </ul>
  
 </div>
  
 %% Cell type:markdown id: tags:
  
 ## Step 8 - Reload data to be sure ;-)
  
 %% Cell type:code id: tags:
  
 ``` python
 pwk.chrono_start()
  
 dataset='set-48x48-L'
 samples=range(24)
  
 with  h5py.File(f'{output_dir}/{dataset}.h5','r') as f:
    x_tmp = f['x_train'][:]
    y_tmp = f['y_train'][:]
    print("dataset loaded from h5 file.")
  
 pwk.plot_images(x_tmp,y_tmp, samples, columns=8, x_size=2, y_size=2,
                colorbar=False, y_pred=None, cm='binary', save_as='16-enhanced_images')
 x_tmp,y_tmp=0,0
  
 pwk.chrono_show()
 ```
  
 %% Output
  
    dataset loaded from h5 file.
  

  
    
    Duration :  00:00:01 801ms
  
 %% Cell type:code id: tags:
  
 ``` python
 pwk.end()
 ```
  
 %% Output
  
    End time is : Monday 11 January 2021, 21:36:05
    Duration is : 00:01:38 239ms
    This notebook ends here
  
 %% Cell type:markdown id: tags:
  
 ---
 <img width="80px" src="../fidle/img/00-Fidle-logo-01.svg"></img>
--- a/GTSRB/01-Preparation-of-data==done==.ipynb
+++ b/GTSRB/01-Preparation-of-data==done==.ipynb
--- a/GTSRB/02-First-convolutions==done==.ipynb
+++ b/GTSRB/02-First-convolutions==done==.ipynb
--- a/GTSRB/03-Tracking-and-visualizing==done==.ipynb
+++ b/GTSRB/03-Tracking-and-visualizing==done==.ipynb
--- a/GTSRB/04-Data-augmentation==done==.ipynb
+++ b/GTSRB/04-Data-augmentation==done==.ipynb
--- a/GTSRB/05-Full-convolutions=1==done==.ipynb
+++ b/GTSRB/05-Full-convolutions=1==done==.ipynb
--- a/GTSRB/05-Full-convolutions=2==done==.ipynb
+++ b/GTSRB/05-Full-convolutions=2==done==.ipynb
--- a/GTSRB/06-Notebook-as-a-batch==done==.ipynb
+++ b/GTSRB/06-Notebook-as-a-batch==done==.ipynb
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "<img width=\"800px\" src=\"../fidle/img/00-Fidle-header-01.svg\"></img>\n",
+    "\n",
+    "# <!-- TITLE --> [GTSRB6] - Full convolutions as a batch\n",
+    "<!-- DESC --> Episode 6 : To compute bigger, use your notebook in batch mode\n",
+    "<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
+    "\n",
+    "## Objectives :\n",
+    "  - Run a notebook code as a **job**\n",
+    "  - Follow up with Tensorboard\n",
+    "  \n",
+    "The German Traffic Sign Recognition Benchmark (GTSRB) is a dataset with more than 50,000 photos of road signs from about 40 classes.  \n",
+    "The final aim is to recognise them !  \n",
+    "Description is available there : http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset\n",
+    "\n",
+    "\n",
+    "## What we're going to do :\n",
+    "Our main steps:\n",
+    " - Run Full-convolution.ipynb as a batch :\n",
+    "    - Notebook mode\n",
+    "    - Script mode \n",
+    " - Tensorboard follow up"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Step 1 - Import and init \n",
+    "Not really useful here ;-)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<style>\n",
+       "\n",
+       "div.warn {    \n",
+       "    background-color: #fcf2f2;\n",
+       "    border-color: #dFb5b4;\n",
+       "    border-left: 5px solid #dfb5b4;\n",
+       "    padding: 0.5em;\n",
+       "    font-weight: bold;\n",
+       "    font-size: 1.1em;;\n",
+       "    }\n",
+       "\n",
+       "\n",
+       "\n",
+       "div.nota {    \n",
+       "    background-color: #DAFFDE;\n",
+       "    border-left: 5px solid #92CC99;\n",
+       "    padding: 0.5em;\n",
+       "    }\n",
+       "\n",
+       "div.todo:before { content:url(data:image/svg+xml;base64,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);\n",
+       "    float:left;\n",
+       "    margin-right:20px;\n",
+       "    margin-top:-20px;\n",
+       "    margin-bottom:20px;\n",
+       "}\n",
+       "div.todo{\n",
+       "    font-weight: bold;\n",
+       "    font-size: 1.1em;\n",
+       "    margin-top:40px;\n",
+       "}\n",
+       "div.todo ul{\n",
+       "    margin: 0.2em;\n",
+       "}\n",
+       "div.todo li{\n",
+       "    margin-left:60px;\n",
+       "    margin-top:0;\n",
+       "    margin-bottom:0;\n",
+       "}\n",
+       "\n",
+       "div .comment{\n",
+       "    font-size:0.8em;\n",
+       "    color:#696969;\n",
+       "}\n",
+       "\n",
+       "\n",
+       "\n",
+       "</style>\n",
+       "\n"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/markdown": [
+       "<br>**FIDLE 2020 - Practical Work Module**"
+      ],
+      "text/plain": [
+       "<IPython.core.display.Markdown object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Version              : 2.0.7\n",
+      "Notebook id          : GTSRB6\n",
+      "Run time             : Wednesday 27 January 2021, 19:11:13\n",
+      "TensorFlow version   : 2.2.0\n",
+      "Keras version        : 2.3.0-tf\n",
+      "Datasets dir         : /gpfswork/rech/mlh/uja62cb/datasets\n",
+      "Run dir              : ./run\n",
+      "Update keras cache   : False\n",
+      "Save figs            : True\n",
+      "Path figs            : ./run/figs\n"
+     ]
+    }
+   ],
+   "source": [
+    "import sys\n",
+    "\n",
+    "sys.path.append('..')\n",
+    "import fidle.pwk as pwk\n",
+    "\n",
+    "datasets_dir = pwk.init('GTSRB6')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Step 2 - How to run a notebook as a batch ?\n",
+    "\n",
+    "Two simple solutions are possible :-)\n",
+    "\n",
+    " - **Option 1 - As a notebook ! (a good choice)**\n",
+    "\n",
+    "  Very simple.  \n",
+    "  The result is the executed notebook, so we can retrieve all the cell'soutputs of the notebook :  \n",
+    "  ```jupyter nbconvert (...) --to notebook --execute <notebook>```   \n",
+    "\n",
+    "  Example :   \n",
+    "  ```jupyter nbconvert --ExecutePreprocessor.timeout=-1 --to notebook --execute my_notebook.ipynb'```  \n",
+    "  The result will be a notebook: 'my_notebook.nbconvert.ipynb'.\n",
+    "  \n",
+    "  See: [nbconvert documentation](https://nbconvert.readthedocs.io/en/latest/usage.html#convert-notebook)\n",
+    "\n",
+    " - **Option 2 - As a script**\n",
+    "\n",
+    "  Very simple too, but with some constraints on the notebook.  \n",
+    "  We will convert the notebook to a Python script (IPython, to be precise) :  \n",
+    "  ```jupyter nbconvert --to script <notebook>```  \n",
+    "  \n",
+    "  Then we can execute this script :  \n",
+    "  ```ipython <script>```\n",
+    "  \n",
+    "  See: [nbconvert documentation](https://nbconvert.readthedocs.io/en/latest/usage.html#executable-script)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Step 3 - Run as a script\n",
+    "\n",
+    "Maybe not always the best solution, but this solution is very rustic !  \n",
+    "\n",
+    "### 3.1 - Convert to IPython script :"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[NbConvertApp] Converting notebook 05-Full-convolutions.ipynb to script\r\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "[NbConvertApp] Writing 12984 bytes to 05-full_convolutions.py\r\n"
+     ]
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "-rw-r--r-- 1 uja62cb mlh 12984 Jan 27 19:11 05-full_convolutions.py\r\n"
+     ]
+    }
+   ],
+   "source": [
+    "! jupyter nbconvert --to script --output='05-full_convolutions' '05-Full-convolutions.ipynb'\n",
+    "! ls -l *.py"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2.2 - Batch submission\n",
+    "\n",
+    "See the two examples of bash launch script :\n",
+    " - `batch_slurm.sh` using Slurm (like at IDRIS)\n",
+    " - `batch_oar.sh`   using OAR (like at GRICAD)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Example at IDRIS\n",
+    "\n",
+    "On the frontal :\n",
+    "```bash\n",
+    "# hostname\n",
+    "jean-zay2\n",
+    "\n",
+    "# sbatch $WORK/fidle/GTSRB/batch_slurm.sh \n",
+    "Submitted batch job 249794\n",
+    "\n",
+    "#squeue -u $USER\n",
+    "             JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)\n",
+    "            249794    gpu_p1 GTSRB Fu  uja62cb PD       0:00      1 (Resources)\n",
+    "\n",
+    "# ls -l _batch/\n",
+    "total 32769\n",
+    "-rw-r--r-- 1 uja62cb gensim01 13349 Sep 10 11:32 GTSRB_249794.err\n",
+    "-rw-r--r-- 1 uja62cb gensim01   489 Sep 10 11:31 GTSRB_249794.out\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### Example at GRICAD\n",
+    "\n",
+    "Have to be done on the frontal :\n",
+    "```bash\n",
+    "# hostname\n",
+    "f-dahu\n",
+    "\n",
+    "# pwd\n",
+    "/home/paroutyj\n",
+    "\n",
+    "# oarsub -S ~/fidle/GTSRB/batch_oar.sh\n",
+    "[GPUNODE] Adding gpu node restriction\n",
+    "[ADMISSION RULE] Modify resource description with type constraints\n",
+    "\n",
+    "#oarstat -u\n",
+    "Job id    S User     Duration   System message\n",
+    "--------- - -------- ---------- ------------------------------------------------\n",
+    "5878410   R paroutyj    0:19:56 R=8,W=1:0:0,J=I,P=fidle,T=gpu (Karma=0.005,quota_ok)\n",
+    "5896266   W paroutyj    0:00:00 R=8,W=1:0:0,J=B,N=Full convolutions,P=fidle,T=gpu\n",
+    "\n",
+    "# ls -l\n",
+    "total 8\n",
+    "-rw-r--r-- 1 paroutyj l-simap    0 Feb 28 15:58 batch_oar_5896266.err\n",
+    "-rw-r--r-- 1 paroutyj l-simap 5703 Feb 28 15:58 batch_oar_5896266.out\n",
+    "```\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "End time is : Wednesday 27 January 2021, 19:11:15\n",
+      "Duration is : 00:00:02 542ms\n",
+      "This notebook ends here\n"
+     ]
+    }
+   ],
+   "source": [
+    "pwk.end()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "----\n",
+    "<div class='todo'>\n",
+    " Your mission if you accept it: Run our full_convolution code in batch mode.<br>\n",
+    " For that :\n",
+    "  <ul>\n",
+    "      <li>Validate the full_convolution notebook on short tests</li>\n",
+    "      <li>Submit it in batch mode for validation</li>\n",
+    "      <li>Modify the notebook for a full run and submit it :-)</li>\n",
+    "  </ul>\n",
+    " \n",
+    "</div>"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "---\n",
+    "<img width=\"80px\" src=\"../fidle/img/00-Fidle-logo-01.svg\"></img>"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.7"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}
+%% Cell type:markdown id: tags:
+
+<img width="800px" src="../fidle/img/00-Fidle-header-01.svg"></img>
+
+# <!-- TITLE --> [GTSRB6] - Full convolutions as a batch
+<!-- DESC --> Episode 6 : To compute bigger, use your notebook in batch mode
+<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->
+
+## Objectives :
+  - Run a notebook code as a **job**
+  - Follow up with Tensorboard
+
+The German Traffic Sign Recognition Benchmark (GTSRB) is a dataset with more than 50,000 photos of road signs from about 40 classes.
+The final aim is to recognise them !
+Description is available there : http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset
+
+
+## What we're going to do :
+Our main steps:
+ - Run Full-convolution.ipynb as a batch :
+    - Notebook mode
+    - Script mode
+ - Tensorboard follow up
+
+%% Cell type:markdown id: tags:
+
+### Step 1 - Import and init
+Not really useful here ;-)
+
+%% Cell type:code id: tags:
+
+``` python
+import sys
+
+sys.path.append('..')
+import fidle.pwk as pwk
+
+datasets_dir = pwk.init('GTSRB6')
+```
+
+%% Output
+
+
+    <br>**FIDLE 2020 - Practical Work Module**
+
+    Version              : 2.0.7
+    Notebook id          : GTSRB6
+    Run time             : Wednesday 27 January 2021, 19:11:13
+    TensorFlow version   : 2.2.0
+    Keras version        : 2.3.0-tf
+    Datasets dir         : /gpfswork/rech/mlh/uja62cb/datasets
+    Run dir              : ./run
+    Update keras cache   : False
+    Save figs            : True
+    Path figs            : ./run/figs
+
+%% Cell type:markdown id: tags:
+
+## Step 2 - How to run a notebook as a batch ?
+
+Two simple solutions are possible :-)
+
+ - **Option 1 - As a notebook ! (a good choice)**
+
+  Very simple.
+  The result is the executed notebook, so we can retrieve all the cell'soutputs of the notebook :
+  ```jupyter nbconvert (...) --to notebook --execute <notebook>```
+
+  Example :
+  ```jupyter nbconvert --ExecutePreprocessor.timeout=-1 --to notebook --execute my_notebook.ipynb'```
+  The result will be a notebook: 'my_notebook.nbconvert.ipynb'.
+
+  See: [nbconvert documentation](https://nbconvert.readthedocs.io/en/latest/usage.html#convert-notebook)
+
+ - **Option 2 - As a script**
+
+  Very simple too, but with some constraints on the notebook.
+  We will convert the notebook to a Python script (IPython, to be precise) :
+  ```jupyter nbconvert --to script <notebook>```
+
+  Then we can execute this script :
+  ```ipython <script>```
+
+  See: [nbconvert documentation](https://nbconvert.readthedocs.io/en/latest/usage.html#executable-script)
+
+%% Cell type:markdown id: tags:
+
+## Step 3 - Run as a script
+
+Maybe not always the best solution, but this solution is very rustic !
+
+### 3.1 - Convert to IPython script :
+
+%% Cell type:code id: tags:
+
+``` python
+! jupyter nbconvert --to script --output='05-full_convolutions' '05-Full-convolutions.ipynb'
+! ls -l *.py
+```
+
+%% Output
+
+    [NbConvertApp] Converting notebook 05-Full-convolutions.ipynb to script
+
+    [NbConvertApp] Writing 12984 bytes to 05-full_convolutions.py
+
+    -rw-r--r-- 1 uja62cb mlh 12984 Jan 27 19:11 05-full_convolutions.py
+
+%% Cell type:markdown id: tags:
+
+### 2.2 - Batch submission
+
+See the two examples of bash launch script :
+ - `batch_slurm.sh` using Slurm (like at IDRIS)
+ - `batch_oar.sh`   using OAR (like at GRICAD)
+
+%% Cell type:markdown id: tags:
+
+#### Example at IDRIS
+
+On the frontal :
+```bash
+# hostname
+jean-zay2
+
+# sbatch $WORK/fidle/GTSRB/batch_slurm.sh
+Submitted batch job 249794
+
+#squeue -u $USER
+             JOBID PARTITION     NAME     USER ST       TIME  NODES NODELIST(REASON)
+            249794    gpu_p1 GTSRB Fu  uja62cb PD       0:00      1 (Resources)
+
+# ls -l _batch/
+total 32769
+-rw-r--r-- 1 uja62cb gensim01 13349 Sep 10 11:32 GTSRB_249794.err
+-rw-r--r-- 1 uja62cb gensim01   489 Sep 10 11:31 GTSRB_249794.out
+```
+
+%% Cell type:markdown id: tags:
+
+#### Example at GRICAD
+
+Have to be done on the frontal :
+```bash
+# hostname
+f-dahu
+
+# pwd
+/home/paroutyj
+
+# oarsub -S ~/fidle/GTSRB/batch_oar.sh
+[GPUNODE] Adding gpu node restriction
+[ADMISSION RULE] Modify resource description with type constraints
+
+#oarstat -u
+Job id    S User     Duration   System message
+--------- - -------- ---------- ------------------------------------------------
+5878410   R paroutyj    0:19:56 R=8,W=1:0:0,J=I,P=fidle,T=gpu (Karma=0.005,quota_ok)
+5896266   W paroutyj    0:00:00 R=8,W=1:0:0,J=B,N=Full convolutions,P=fidle,T=gpu
+
+# ls -l
+total 8
+-rw-r--r-- 1 paroutyj l-simap    0 Feb 28 15:58 batch_oar_5896266.err
+-rw-r--r-- 1 paroutyj l-simap 5703 Feb 28 15:58 batch_oar_5896266.out
+```
+
+%% Cell type:code id: tags:
+
+``` python
+pwk.end()
+```
+
+%% Output
+
+    End time is : Wednesday 27 January 2021, 19:11:15
+    Duration is : 00:00:02 542ms
+    This notebook ends here
+
+%% Cell type:markdown id: tags:
+
+----
+<div class='todo'>
+ Your mission if you accept it: Run our full_convolution code in batch mode.<br>
+ For that :
+  <ul>
+      <li>Validate the full_convolution notebook on short tests</li>
+      <li>Submit it in batch mode for validation</li>
+      <li>Modify the notebook for a full run and submit it :-)</li>
+  </ul>
+
+</div>
+
+%% Cell type:markdown id: tags:
+
+---
+<img width="80px" src="../fidle/img/00-Fidle-logo-01.svg"></img>
--- a/GTSRB/07-Show-report==done==.ipynb
+++ b/GTSRB/07-Show-report==done==.ipynb
--- a/IMDB/01-Embedding-Keras==done==.ipynb
+++ b/IMDB/01-Embedding-Keras==done==.ipynb
--- a/IMDB/02-Prediction==done==.ipynb
+++ b/IMDB/02-Prediction==done==.ipynb
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "<img width=\"800px\" src=\"../fidle/img/00-Fidle-header-01.svg\"></img>\n",
+    "\n",
+    "# <!-- TITLE --> [IMDB2] - Reload and reuse a saved model\n",
+    "<!-- DESC --> Retrieving a saved model to perform a sentiment analysis (movie review)\n",
+    "<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
+    "\n",
+    "## Objectives :\n",
+    " - The objective is to guess whether film reviews are **positive or negative** based on the analysis of the text. \n",
+    " - For this, we will use our **previously saved model**.\n",
+    "\n",
+    "Original dataset can be find **[there](http://ai.stanford.edu/~amaas/data/sentiment/)**  \n",
+    "Note that [IMDb.com](https://imdb.com) offers several easy-to-use [datasets](https://www.imdb.com/interfaces/)  \n",
+    "For simplicity's sake, we'll use the dataset directly [embedded in Keras](https://www.tensorflow.org/api_docs/python/tf/keras/datasets)\n",
+    "\n",
+    "## What we're going to do :\n",
+    "\n",
+    " - Preparing the data\n",
+    " - Retrieve our saved model\n",
+    " - Evaluate the result\n"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Step 1 - Init python stuff"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/html": [
+       "<style>\n",
+       "\n",
+       "div.warn {    \n",
+       "    background-color: #fcf2f2;\n",
+       "    border-color: #dFb5b4;\n",
+       "    border-left: 5px solid #dfb5b4;\n",
+       "    padding: 0.5em;\n",
+       "    font-weight: bold;\n",
+       "    font-size: 1.1em;;\n",
+       "    }\n",
+       "\n",
+       "\n",
+       "\n",
+       "div.nota {    \n",
+       "    background-color: #DAFFDE;\n",
+       "    border-left: 5px solid #92CC99;\n",
+       "    padding: 0.5em;\n",
+       "    }\n",
+       "\n",
+       "div.todo:before { content:url(data:image/svg+xml;base64,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);\n",
+       "    float:left;\n",
+       "    margin-right:20px;\n",
+       "    margin-top:-20px;\n",
+       "    margin-bottom:20px;\n",
+       "}\n",
+       "div.todo{\n",
+       "    font-weight: bold;\n",
+       "    font-size: 1.1em;\n",
+       "    margin-top:40px;\n",
+       "}\n",
+       "div.todo ul{\n",
+       "    margin: 0.2em;\n",
+       "}\n",
+       "div.todo li{\n",
+       "    margin-left:60px;\n",
+       "    margin-top:0;\n",
+       "    margin-bottom:0;\n",
+       "}\n",
+       "\n",
+       "div .comment{\n",
+       "    font-size:0.8em;\n",
+       "    color:#696969;\n",
+       "}\n",
+       "\n",
+       "\n",
+       "\n",
+       "</style>\n",
+       "\n"
+      ],
+      "text/plain": [
+       "<IPython.core.display.HTML object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "data": {
+      "text/markdown": [
+       "<br>**FIDLE 2020 - Practical Work Module**"
+      ],
+      "text/plain": [
+       "<IPython.core.display.Markdown object>"
+      ]
+     },
+     "metadata": {},
+     "output_type": "display_data"
+    },
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "Version              : 2.0.7\n",
+      "Notebook id          : IMDB2\n",
+      "Run time             : Wednesday 27 January 2021, 19:12:12\n",
+      "TensorFlow version   : 2.2.0\n",
+      "Keras version        : 2.3.0-tf\n",
+      "Datasets dir         : /gpfswork/rech/mlh/uja62cb/datasets\n",
+      "Run dir              : ./run\n",
+      "Update keras cache   : False\n",
+      "Save figs            : True\n",
+      "Path figs            : ./run/figs\n"
+     ]
+    }
+   ],
+   "source": [
+    "import numpy as np\n",
+    "\n",
+    "import tensorflow as tf\n",
+    "import tensorflow.keras as keras\n",
+    "import tensorflow.keras.datasets.imdb as imdb\n",
+    "\n",
+    "import matplotlib.pyplot as plt\n",
+    "import matplotlib\n",
+    "import pandas as pd\n",
+    "\n",
+    "import os,sys,h5py,json,re\n",
+    "\n",
+    "from importlib import reload\n",
+    "\n",
+    "sys.path.append('..')\n",
+    "import fidle.pwk as pwk\n",
+    "\n",
+    "datasets_dir = pwk.init('IMDB2')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Step 2 : Preparing the data\n",
+    "### 2.1 - Our reviews :"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "reviews = [ \"This film is particularly nice, a must see.\",\n",
+    "             \"Some films are great classics and cannot be ignored.\",\n",
+    "             \"This movie is just abominable and doesn't deserve to be seen!\"]"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2.2 - Retrieve dictionaries\n",
+    "Note : This dictionary is generated by [01-Embedding-Keras](01-Embedding-Keras.ipynb) notebook."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open('./data/word_index.json', 'r') as fp:\n",
+    "    word_index = json.load(fp)\n",
+    "    index_word = {index:word for word,index in word_index.items()} "
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2.3 - Clean, index and padd"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "max_len    = 256\n",
+    "vocab_size = 10000\n",
+    "\n",
+    "\n",
+    "nb_reviews = len(reviews)\n",
+    "x_data     = []\n",
+    "\n",
+    "# ---- For all reviews\n",
+    "for review in reviews:\n",
+    "    # ---- First index must be <start>\n",
+    "    index_review=[1]\n",
+    "    # ---- For all words\n",
+    "    for w in review.split(' '):\n",
+    "        # ---- Clean it\n",
+    "        w_clean = re.sub(r\"[^a-zA-Z0-9]\", \"\", w)\n",
+    "        # ---- Not empty ?\n",
+    "        if len(w_clean)>0:\n",
+    "            # ---- Get the index\n",
+    "            w_index = word_index.get(w,2)\n",
+    "            if w_index>vocab_size : w_index=2\n",
+    "            # ---- Add the index if < vocab_size\n",
+    "            index_review.append(w_index)\n",
+    "    # ---- Add the indexed review\n",
+    "    x_data.append(index_review)    \n",
+    "\n",
+    "# ---- Padding\n",
+    "x_data = keras.preprocessing.sequence.pad_sequences(x_data, value   = 0, padding = 'post', maxlen  = max_len)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### 2.4 - Have a look"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "Text review      : This film is particularly nice, a must see.\n",
+      "x_train[0]       : [1, 2, 22, 9, 572, 2, 6, 215, 2, 0, 0, 0, 0, 0] (...)\n",
+      "Translation      : <start> <unknown> film is particularly <unknown> a must <unknown> <pad> <pad> <pad> <pad> <pad> (...)\n",
+      "\n",
+      "Text review      : Some films are great classics and cannot be ignored.\n",
+      "x_train[1]       : [1, 2, 108, 26, 87, 2239, 5, 566, 30, 2, 0, 0, 0, 0, 0] (...)\n",
+      "Translation      : <start> <unknown> films are great classics and cannot be <unknown> <pad> <pad> <pad> <pad> <pad> (...)\n",
+      "\n",
+      "Text review      : This movie is just abominable and doesn't deserve to be seen!\n",
+      "x_train[2]       : [1, 2, 20, 9, 43, 2, 5, 152, 1833, 8, 30, 2, 0, 0, 0, 0, 0] (...)\n",
+      "Translation      : <start> <unknown> movie is just <unknown> and doesn't deserve to be <unknown> <pad> <pad> <pad> <pad> <pad> (...)\n"
+     ]
+    }
+   ],
+   "source": [
+    "def translate(x):\n",
+    "    return ' '.join( [index_word.get(i,'?') for i in x] )\n",
+    "\n",
+    "for i in range(nb_reviews):\n",
+    "    imax=np.where(x_data[i]==0)[0][0]+5\n",
+    "    print(f'\\nText review      :',    reviews[i])\n",
+    "    print(  f'x_train[{i:}]       :', list(x_data[i][:imax]), '(...)')\n",
+    "    print(  'Translation      :', translate(x_data[i][:imax]), '(...)')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Step 2 - Bring back the model"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model = keras.models.load_model('./run/models/best_model.h5')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## Step 4 - Predict"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 7,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "y_pred   = model.predict(x_data)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "#### And the winner is :"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 8,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "\n",
+      "This film is particularly nice, a must see.                            => POSITIVE (0.56)\n",
+      "\n",
+      "Some films are great classics and cannot be ignored.                   => POSITIVE (0.62)\n",
+      "\n",
+      "This movie is just abominable and doesn't deserve to be seen!          => NEGATIVE (0.34)\n"
+     ]
+    }
+   ],
+   "source": [
+    "for i in range(nb_reviews):\n",
+    "    print(f'\\n{reviews[i]:<70} =>',('NEGATIVE' if y_pred[i][0]<0.5 else 'POSITIVE'),f'({y_pred[i][0]:.2f})')"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 9,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "End time is : Wednesday 27 January 2021, 19:12:14\n",
+      "Duration is : 00:00:02 516ms\n",
+      "This notebook ends here\n"
+     ]
+    }
+   ],
+   "source": [
+    "pwk.end()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "---\n",
+    "<img width=\"80px\" src=\"../fidle/img/00-Fidle-logo-01.svg\"></img>"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.7"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}
+%% Cell type:markdown id: tags:
+
+<img width="800px" src="../fidle/img/00-Fidle-header-01.svg"></img>
+
+# <!-- TITLE --> [IMDB2] - Reload and reuse a saved model
+<!-- DESC --> Retrieving a saved model to perform a sentiment analysis (movie review)
+<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->
+
+## Objectives :
+ - The objective is to guess whether film reviews are **positive or negative** based on the analysis of the text.
+ - For this, we will use our **previously saved model**.
+
+Original dataset can be find **[there](http://ai.stanford.edu/~amaas/data/sentiment/)**
+Note that [IMDb.com](https://imdb.com) offers several easy-to-use [datasets](https://www.imdb.com/interfaces/)
+For simplicity's sake, we'll use the dataset directly [embedded in Keras](https://www.tensorflow.org/api_docs/python/tf/keras/datasets)
+
+## What we're going to do :
+
+ - Preparing the data
+ - Retrieve our saved model
+ - Evaluate the result
+
+%% Cell type:markdown id: tags:
+
+## Step 1 - Init python stuff
+
+%% Cell type:code id: tags:
+
+``` python
+import numpy as np
+
+import tensorflow as tf
+import tensorflow.keras as keras
+import tensorflow.keras.datasets.imdb as imdb
+
+import matplotlib.pyplot as plt
+import matplotlib
+import pandas as pd
+
+import os,sys,h5py,json,re
+
+from importlib import reload
+
+sys.path.append('..')
+import fidle.pwk as pwk
+
+datasets_dir = pwk.init('IMDB2')
+```
+
+%% Output
+
+
+    <br>**FIDLE 2020 - Practical Work Module**
+
+    Version              : 2.0.7
+    Notebook id          : IMDB2
+    Run time             : Wednesday 27 January 2021, 19:12:12
+    TensorFlow version   : 2.2.0
+    Keras version        : 2.3.0-tf
+    Datasets dir         : /gpfswork/rech/mlh/uja62cb/datasets
+    Run dir              : ./run
+    Update keras cache   : False
+    Save figs            : True
+    Path figs            : ./run/figs
+
+%% Cell type:markdown id: tags:
+
+## Step 2 : Preparing the data
+### 2.1 - Our reviews :
+
+%% Cell type:code id: tags:
+
+``` python
+reviews = [ "This film is particularly nice, a must see.",
+             "Some films are great classics and cannot be ignored.",
+             "This movie is just abominable and doesn't deserve to be seen!"]
+```
+
+%% Cell type:markdown id: tags:
+
+### 2.2 - Retrieve dictionaries
+Note : This dictionary is generated by [01-Embedding-Keras](01-Embedding-Keras.ipynb) notebook.
+
+%% Cell type:code id: tags:
+
+``` python
+with open('./data/word_index.json', 'r') as fp:
+    word_index = json.load(fp)
+    index_word = {index:word for word,index in word_index.items()}
+```
+
+%% Cell type:markdown id: tags:
+
+### 2.3 - Clean, index and padd
+
+%% Cell type:code id: tags:
+
+``` python
+max_len    = 256
+vocab_size = 10000
+
+
+nb_reviews = len(reviews)
+x_data     = []
+
+# ---- For all reviews
+for review in reviews:
+    # ---- First index must be <start>
+    index_review=[1]
+    # ---- For all words
+    for w in review.split(' '):
+        # ---- Clean it
+        w_clean = re.sub(r"[^a-zA-Z0-9]", "", w)
+        # ---- Not empty ?
+        if len(w_clean)>0:
+            # ---- Get the index
+            w_index = word_index.get(w,2)
+            if w_index>vocab_size : w_index=2
+            # ---- Add the index if < vocab_size
+            index_review.append(w_index)
+    # ---- Add the indexed review
+    x_data.append(index_review)
+
+# ---- Padding
+x_data = keras.preprocessing.sequence.pad_sequences(x_data, value   = 0, padding = 'post', maxlen  = max_len)
+```
+
+%% Cell type:markdown id: tags:
+
+### 2.4 - Have a look
+
+%% Cell type:code id: tags:
+
+``` python
+def translate(x):
+    return ' '.join( [index_word.get(i,'?') for i in x] )
+
+for i in range(nb_reviews):
+    imax=np.where(x_data[i]==0)[0][0]+5
+    print(f'\nText review      :',    reviews[i])
+    print(  f'x_train[{i:}]       :', list(x_data[i][:imax]), '(...)')
+    print(  'Translation      :', translate(x_data[i][:imax]), '(...)')
+```
+
+%% Output
+
+    
+    Text review      : This film is particularly nice, a must see.
+    x_train[0]       : [1, 2, 22, 9, 572, 2, 6, 215, 2, 0, 0, 0, 0, 0] (...)
+    Translation      : <start> <unknown> film is particularly <unknown> a must <unknown> <pad> <pad> <pad> <pad> <pad> (...)
+    
+    Text review      : Some films are great classics and cannot be ignored.
+    x_train[1]       : [1, 2, 108, 26, 87, 2239, 5, 566, 30, 2, 0, 0, 0, 0, 0] (...)
+    Translation      : <start> <unknown> films are great classics and cannot be <unknown> <pad> <pad> <pad> <pad> <pad> (...)
+    
+    Text review      : This movie is just abominable and doesn't deserve to be seen!
+    x_train[2]       : [1, 2, 20, 9, 43, 2, 5, 152, 1833, 8, 30, 2, 0, 0, 0, 0, 0] (...)
+    Translation      : <start> <unknown> movie is just <unknown> and doesn't deserve to be <unknown> <pad> <pad> <pad> <pad> <pad> (...)
+
+%% Cell type:markdown id: tags:
+
+## Step 2 - Bring back the model
+
+%% Cell type:code id: tags:
+
+``` python
+model = keras.models.load_model('./run/models/best_model.h5')
+```
+
+%% Cell type:markdown id: tags:
+
+## Step 4 - Predict
+
+%% Cell type:code id: tags:
+
+``` python
+y_pred   = model.predict(x_data)
+```
+
+%% Cell type:markdown id: tags:
+
+#### And the winner is :
+
+%% Cell type:code id: tags:
+
+``` python
+for i in range(nb_reviews):
+    print(f'\n{reviews[i]:<70} =>',('NEGATIVE' if y_pred[i][0]<0.5 else 'POSITIVE'),f'({y_pred[i][0]:.2f})')
+```
+
+%% Output
+
+    
+    This film is particularly nice, a must see.                            => POSITIVE (0.56)
+    
+    Some films are great classics and cannot be ignored.                   => POSITIVE (0.62)
+    
+    This movie is just abominable and doesn't deserve to be seen!          => NEGATIVE (0.34)
+
+%% Cell type:code id: tags:
+
+``` python
+pwk.end()
+```
+
+%% Output
+
+    End time is : Wednesday 27 January 2021, 19:12:14
+    Duration is : 00:00:02 516ms
+    This notebook ends here
+
+%% Cell type:markdown id: tags:
+
+---
+<img width="80px" src="../fidle/img/00-Fidle-logo-01.svg"></img>
--- a/IMDB/03-LSTM-Keras==done==.ipynb
+++ b/IMDB/03-LSTM-Keras==done==.ipynb
--- a/IRIS/01-Simple-Perceptron==done==.ipynb
+++ b/IRIS/01-Simple-Perceptron==done==.ipynb
--- a/LinearReg/01-Linear-Regression==done==.ipynb
+++ b/LinearReg/01-Linear-Regression==done==.ipynb
--- a/LinearReg/02-Gradient-descent==done==.ipynb
+++ b/LinearReg/02-Gradient-descent==done==.ipynb
--- a/LinearReg/03-Polynomial-Regression==done==.ipynb
+++ b/LinearReg/03-Polynomial-Regression==done==.ipynb