diff --git a/BHPD.Keras3/02-DNN-Regression-Premium.ipynb b/BHPD.Keras3/02-DNN-Regression-Premium.ipynb
index d3a23c48895fe7ad95dea25c8dec4712a53877ce..65cdd8bd312dffde3374e99ee2fa463427ad70fa 100644
--- a/BHPD.Keras3/02-DNN-Regression-Premium.ipynb
+++ b/BHPD.Keras3/02-DNN-Regression-Premium.ipynb
@@ -284,8 +284,8 @@
"metadata": {},
"outputs": [],
"source": [
- "os.makedirs('./run/models', mode=0o750, exist_ok=True)\n",
- "save_dir = \"./run/models/best_model.keras\"\n",
+ "os.makedirs(f'{run_dir}/models', mode=0o750, exist_ok=True)\n",
+ "save_dir = f'{run_dir}/models/best_model.keras'\n",
"\n",
"savemodel_callback = keras.callbacks.ModelCheckpoint( filepath=save_dir, monitor='val_mae', mode='max', save_best_only=True)"
]
@@ -383,7 +383,7 @@
"metadata": {},
"outputs": [],
"source": [
- "loaded_model = keras.models.load_model('./run/models/best_model.keras')\n",
+ "loaded_model = keras.models.load_model(f'{run_dir}/models/best_model.keras')\n",
"loaded_model.summary()\n",
"print(\"Loaded.\")"
]
diff --git a/Embedding.Keras2/01-One-hot-encoding.ipynb b/Embedding.Keras3/01-One-hot-encoding.ipynb
similarity index 87%
rename from Embedding.Keras2/01-One-hot-encoding.ipynb
rename to Embedding.Keras3/01-One-hot-encoding.ipynb
index e4e689f847225906dce47c80089d8af93708ed55..7c7451c72d7da7a9a014506a04ec7c8cfeb07fdf 100644
--- a/Embedding.Keras2/01-One-hot-encoding.ipynb
+++ b/Embedding.Keras3/01-One-hot-encoding.ipynb
@@ -6,8 +6,8 @@
"source": [
"<img width=\"800px\" src=\"../fidle/img/header.svg\"></img>\n",
"\n",
- "# <!-- TITLE --> [K2IMDB1] - Sentiment analysis with hot-one encoding\n",
- "<!-- DESC --> A basic example of sentiment analysis with sparse encoding, using a dataset from Internet Movie Database (IMDB), using Keras 2 and Tensorflow (obsolete)\n",
+ "# <!-- TITLE --> [K3IMDB1] - Sentiment analysis with hot-one encoding\n",
+ "<!-- DESC --> A basic example of sentiment analysis with sparse encoding, using a dataset from Internet Movie Database (IMDB), using Keras 3 on PyTorch\n",
"<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
"\n",
"## Objectives :\n",
@@ -41,24 +41,20 @@
"metadata": {},
"outputs": [],
"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",
+ "import os\n",
+ "os.environ['KERAS_BACKEND'] = 'torch'\n",
"\n",
- "import matplotlib.pyplot as plt\n",
- "import matplotlib\n",
+ "import keras\n",
+ "import keras.datasets.imdb as imdb\n",
"\n",
+ "import numpy as np\n",
"import pandas as pd\n",
- "\n",
- "import os,sys,h5py,json\n",
- "from importlib import reload\n",
+ "import matplotlib.pyplot as plt\n",
"\n",
"import fidle\n",
"\n",
"# Init Fidle environment\n",
- "run_id, run_dir, datasets_dir = fidle.init('K2IMDB1')"
+ "run_id, run_dir, datasets_dir = fidle.init('K3IMDB1')"
]
},
{
@@ -232,17 +228,22 @@
"source": [
"# ----- Retrieve x,y\n",
"#\n",
- "(x_train, y_train), (x_test, y_test) = imdb.load_data( num_words=vocab_size, skip_top=hide_most_frequently)\n",
+ "start_char = 1 # Start of a sequence (padding is 0)\n",
+ "oov_char = 2 # Out-of-vocabulary\n",
+ "index_from = 3 # First word id\n",
"\n",
- "y_train = np.asarray(y_train).astype('float32')\n",
- "y_test = np.asarray(y_test ).astype('float32')\n",
+ "(x_train, y_train), (x_test, y_test) = imdb.load_data( num_words = vocab_size, \n",
+ " skip_top = hide_most_frequently,\n",
+ " start_char = start_char, \n",
+ " oov_char = oov_char, \n",
+ " index_from = index_from)\n",
"\n",
"# ---- About\n",
"#\n",
"print(\"Max(x_train,x_test) : \", fidle.utils.rmax([x_train,x_test]) )\n",
"print(\"Min(x_train,x_test) : \", fidle.utils.rmin([x_train,x_test]) )\n",
- "print(\"x_train : {} y_train : {}\".format(x_train.shape, y_train.shape))\n",
- "print(\"x_test : {} y_test : {}\".format(x_test.shape, y_test.shape))"
+ "print(\"Len(x_train) : \", len(x_train))\n",
+ "print(\"Len(x_test) : \", len(x_test))\n"
]
},
{
@@ -283,9 +284,9 @@
"#\n",
"word_index = imdb.get_word_index()\n",
"\n",
- "# ---- Shift the dictionary from +3\n",
+ "# ---- Shift the dictionary from <index_from>\n",
"#\n",
- "word_index = {w:(i+3) for w,i in word_index.items()}\n",
+ "word_index = {w:(i+index_from) for w,i in word_index.items()}\n",
"\n",
"# ---- Add <pad>, <start> and <unknown> tags\n",
"#\n",
@@ -340,7 +341,7 @@
"outputs": [],
"source": [
"sizes=[len(i) for i in x_train]\n",
- "plt.figure(figsize=(16,6))\n",
+ "plt.figure(figsize=(12,4))\n",
"plt.hist(sizes, bins=400)\n",
"plt.gca().set(title='Distribution of reviews by size - [{:5.2f}, {:5.2f}]'.format(min(sizes),max(sizes)), \n",
" xlabel='Size', ylabel='Density', xlim=[0,1500])\n",
@@ -354,8 +355,8 @@
"metadata": {},
"outputs": [],
"source": [
- "unk=[ 100*(s.count(2)/len(s)) for s in x_train]\n",
- "plt.figure(figsize=(16,6))\n",
+ "unk=[ 100*(s.count(oov_char)/len(s)) for s in x_train]\n",
+ "plt.figure(figsize=(12,4))\n",
"plt.hist(unk, bins=100)\n",
"plt.gca().set(title='Percent of unknown words - [{:5.2f}, {:5.2f}]'.format(min(unk),max(unk)), \n",
" xlabel='# unknown', ylabel='Density', xlim=[0,30])\n",
@@ -384,7 +385,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 5.1 - Our one-hot encoder"
+ "### 5.1 - Our one-hot encoder function"
]
},
{
@@ -394,7 +395,7 @@
"outputs": [],
"source": [
"def one_hot_encoder(x, vector_size=10000):\n",
- " \n",
+ "\n",
" # ---- Set all to 0\n",
" #\n",
" x_encoded = np.zeros((len(x), vector_size))\n",
@@ -431,7 +432,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "## Step 6 - Build the model"
+ "## Step 6 - Build a nice model"
]
},
{
@@ -440,35 +441,16 @@
"metadata": {},
"outputs": [],
"source": [
- "def get_model(vector_size=10000):\n",
- " \n",
- " model = keras.Sequential()\n",
- " model.add(keras.layers.Input( shape=(vector_size,) ))\n",
- " model.add(keras.layers.Dense( 32, activation='relu'))\n",
- " model.add(keras.layers.Dense( 32, activation='relu'))\n",
- " model.add(keras.layers.Dense( 1, activation='sigmoid'))\n",
+ "model = keras.Sequential(name='My IMDB classifier')\n",
+ "\n",
+ "model.add(keras.layers.Input( shape=(vocab_size,) ))\n",
+ "model.add(keras.layers.Dense( 32, activation='relu'))\n",
+ "model.add(keras.layers.Dense( 32, activation='relu'))\n",
+ "model.add(keras.layers.Dense( 1, activation='sigmoid'))\n",
" \n",
- " model.compile(optimizer = 'rmsprop',\n",
+ "model.compile(optimizer = 'rmsprop',\n",
" loss = 'binary_crossentropy',\n",
" metrics = ['accuracy'])\n",
- " return model"
- ]
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "## Step 7 - Train the model\n",
- "### 7.1 - Get it"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": [
- "model = get_model(vector_size=vocab_size)\n",
"\n",
"model.summary()"
]
@@ -477,7 +459,8 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 7.2 - Add callback"
+ "## Step 7 - Train the model\n",
+ "### 7.1 - Add callback"
]
},
{
@@ -487,15 +470,16 @@
"outputs": [],
"source": [
"os.makedirs(f'{run_dir}/models', mode=0o750, exist_ok=True)\n",
- "save_dir = f'{run_dir}/models/best_model.h5'\n",
- "savemodel_callback = tf.keras.callbacks.ModelCheckpoint(filepath=save_dir, verbose=0, save_best_only=True)"
+ "save_dir = f'{run_dir}/models/best_model.keras'\n",
+ "\n",
+ "savemodel_callback = keras.callbacks.ModelCheckpoint( filepath=save_dir, monitor='val_accuracy', mode='max', save_best_only=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 7.3 - Train it"
+ "### 7.2 - Train it"
]
},
{
@@ -545,13 +529,14 @@
"metadata": {},
"outputs": [],
"source": [
- "model = keras.models.load_model(f'{run_dir}/models/best_model.h5')\n",
+ "model = keras.models.load_model(f'{run_dir}/models/best_model.keras')\n",
"\n",
"# ---- Evaluate\n",
"score = model.evaluate(x_test, y_test, verbose=0)\n",
"\n",
- "print('x_test / loss : {:5.4f}'.format(score[0]))\n",
- "print('x_test / accuracy : {:5.4f}'.format(score[1]))\n",
+ "print('\\n\\nModel evaluation :\\n')\n",
+ "print(' x_test / loss : {:5.4f}'.format(score[0]))\n",
+ "print(' x_test / accuracy : {:5.4f}'.format(score[1]))\n",
"\n",
"values=[score[1], 1-score[1]]\n",
"fidle.scrawler.donut(values,[\"Accuracy\",\"Errors\"], title=\"#### Accuracy donut is :\", save_as='03-donut')\n",
diff --git a/Embedding.Keras2/02-Keras-embedding.ipynb b/Embedding.Keras3/02-Keras-embedding.ipynb
similarity index 81%
rename from Embedding.Keras2/02-Keras-embedding.ipynb
rename to Embedding.Keras3/02-Keras-embedding.ipynb
index b70f5a33d36864c01a32d101f0bccb4aab0ff3a8..21f5fa001b4a4e3d4ea75141a6170580644aaa16 100644
--- a/Embedding.Keras2/02-Keras-embedding.ipynb
+++ b/Embedding.Keras3/02-Keras-embedding.ipynb
@@ -6,8 +6,8 @@
"source": [
"<img width=\"800px\" src=\"../fidle/img/header.svg\"></img>\n",
"\n",
- "# <!-- TITLE --> [K2IMDB2] - Sentiment analysis with text embedding\n",
- "<!-- DESC --> A very classical example of word embedding with a dataset from Internet Movie Database (IMDB), using Keras 2 and Tensorflow (obsolete)\n",
+ "# <!-- TITLE --> [K3IMDB2] - Sentiment analysis with text embedding\n",
+ "<!-- DESC --> A very classical example of word embedding with a dataset from Internet Movie Database (IMDB), using Keras 3 on PyTorch\n",
"<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
"\n",
"## Objectives :\n",
@@ -41,22 +41,21 @@
"metadata": {},
"outputs": [],
"source": [
- "import numpy as np\n",
+ "import os\n",
+ "os.environ['KERAS_BACKEND'] = 'torch'\n",
"\n",
- "import tensorflow as tf\n",
- "import tensorflow.keras as keras\n",
- "import tensorflow.keras.datasets.imdb as imdb\n",
+ "import keras\n",
+ "import keras.datasets.imdb as imdb\n",
"\n",
+ "import h5py,json\n",
+ "import numpy as np\n",
+ "import pandas as pd\n",
"import matplotlib.pyplot as plt\n",
- "import matplotlib\n",
- "\n",
- "import os,sys,h5py,json\n",
- "from importlib import reload\n",
"\n",
"import fidle\n",
"\n",
"# Init Fidle environment\n",
- "run_id, run_dir, datasets_dir = fidle.init('K2IMDB2')"
+ "run_id, run_dir, datasets_dir = fidle.init('K3IMDB2')"
]
},
{
@@ -134,17 +133,24 @@
"metadata": {},
"outputs": [],
"source": [
- "(x_train, y_train), (x_test, y_test) = imdb.load_data( num_words=vocab_size, skip_top=hide_most_frequently, seed= 42,)\n",
+ "# ----- Retrieve x,y\n",
+ "#\n",
+ "start_char = 1 # Start of a sequence (padding is 0)\n",
+ "oov_char = 2 # Out-of-vocabulary\n",
+ "index_from = 3 # First word id\n",
"\n",
- "y_train = np.asarray(y_train).astype('float32')\n",
- "y_test = np.asarray(y_test ).astype('float32')\n",
+ "(x_train, y_train), (x_test, y_test) = imdb.load_data( num_words = vocab_size, \n",
+ " skip_top = hide_most_frequently,\n",
+ " start_char = start_char, \n",
+ " oov_char = oov_char, \n",
+ " index_from = index_from)\n",
"\n",
"# ---- About\n",
"#\n",
"print(\"Max(x_train,x_test) : \", fidle.utils.rmax([x_train,x_test]) )\n",
"print(\"Min(x_train,x_test) : \", fidle.utils.rmin([x_train,x_test]) )\n",
- "print(\"x_train : {} y_train : {}\".format(x_train.shape, y_train.shape))\n",
- "print(\"x_test : {} y_test : {}\".format(x_test.shape, y_test.shape))"
+ "print(\"Len(x_train) : \", len(x_train))\n",
+ "print(\"Len(x_test) : \", len(x_test))"
]
},
{
@@ -167,7 +173,7 @@
"# Create a reverse dictionary : {index:word}\n",
"#\n",
"word_index = imdb.get_word_index()\n",
- "word_index = {w:(i+3) for w,i in word_index.items()}\n",
+ "word_index = {w:(i+index_from) for w,i in word_index.items()}\n",
"word_index.update( {'<pad>':0, '<start>':1, '<unknown>':2, '<undef>':3,} )\n",
"index_word = {index:word for word,index in word_index.items()} \n",
"\n",
@@ -184,7 +190,7 @@
"## Step 3 - Preprocess the data (padding)\n",
"In order to be processed by an NN, all entries must have the **same length.** \n",
"We chose a review length of **review_len** \n",
- "We will therefore complete them with a padding (of \\<pad\\>\\) "
+ "We will therefore complete them with a padding (of 0 as \\<pad\\>\\) "
]
},
{
@@ -229,14 +235,11 @@
" f.create_dataset(\"y_train\", data=y_train)\n",
" f.create_dataset(\"x_test\", data=x_test)\n",
" f.create_dataset(\"y_test\", data=y_test)\n",
+ " print('Dataset h5 file saved.')\n",
"\n",
"with open(f'{output_dir}/word_index.json', 'w') as fp:\n",
" json.dump(word_index, fp)\n",
- "\n",
- "with open(f'{output_dir}/index_word.json', 'w') as fp:\n",
- " json.dump(index_word, fp)\n",
- "\n",
- "print('Saved.')"
+ " print('Word to index saved.')"
]
},
{
@@ -256,38 +259,18 @@
"metadata": {},
"outputs": [],
"source": [
- "def get_model(vocab_size=10000, dense_vector_size=32, review_len=256):\n",
- " \n",
- " model = keras.Sequential()\n",
- " model.add(keras.layers.Input( shape=(review_len,) ))\n",
- " model.add(keras.layers.Embedding(input_dim = vocab_size, \n",
- " output_dim = dense_vector_size, \n",
- " input_length = review_len))\n",
- " model.add(keras.layers.GlobalAveragePooling1D())\n",
- " model.add(keras.layers.Dense(dense_vector_size, activation='relu'))\n",
- " model.add(keras.layers.Dense(1, activation='sigmoid'))\n",
- "\n",
- " model.compile(optimizer = 'adam',\n",
- " loss = 'binary_crossentropy',\n",
- " metrics = ['accuracy'])\n",
- " return model"
- ]
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "## Step 5 - Train the model\n",
- "### 5.1 - Get it"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": [
- "model = get_model(vocab_size, dense_vector_size, review_len)\n",
+ "model = keras.Sequential(name='Embedding model')\n",
+ "\n",
+ "model.add(keras.layers.Input( shape=(review_len,) ))\n",
+ "model.add(keras.layers.Embedding( input_dim = vocab_size,\n",
+ " output_dim = dense_vector_size))\n",
+ "model.add(keras.layers.GlobalAveragePooling1D())\n",
+ "model.add(keras.layers.Dense(dense_vector_size, activation='relu'))\n",
+ "model.add(keras.layers.Dense(1, activation='sigmoid'))\n",
+ "\n",
+ "model.compile( optimizer = 'adam',\n",
+ " loss = 'binary_crossentropy',\n",
+ " metrics = ['accuracy'])\n",
"\n",
"model.summary()"
]
@@ -296,7 +279,8 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 5.2 - Add callback"
+ "## Step 5 - Train the model\n",
+ "### 5.1 Add Callbacks"
]
},
{
@@ -306,15 +290,16 @@
"outputs": [],
"source": [
"os.makedirs(f'{run_dir}/models', mode=0o750, exist_ok=True)\n",
- "save_dir = f'{run_dir}/models/best_model.h5'\n",
- "savemodel_callback = tf.keras.callbacks.ModelCheckpoint(filepath=save_dir, verbose=0, save_best_only=True)"
+ "save_dir = f'{run_dir}/models/best_model.keras'\n",
+ "\n",
+ "savemodel_callback = keras.callbacks.ModelCheckpoint( filepath=save_dir, monitor='val_accuracy', mode='max', save_best_only=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 5.1 - Train it"
+ "### 5.2 - Train it"
]
},
{
@@ -364,7 +349,7 @@
"metadata": {},
"outputs": [],
"source": [
- "model = keras.models.load_model(f'{run_dir}/models/best_model.h5')\n",
+ "model = keras.models.load_model(f'{run_dir}/models/best_model.keras')\n",
"\n",
"# ---- Evaluate\n",
"score = model.evaluate(x_test, y_test, verbose=0)\n",
diff --git a/Embedding.Keras2/03-Prediction.ipynb b/Embedding.Keras3/03-Prediction.ipynb
similarity index 84%
rename from Embedding.Keras2/03-Prediction.ipynb
rename to Embedding.Keras3/03-Prediction.ipynb
index f058e29a885f4dcbf8e50c3618fbfed646e4db14..ed1642084b1b904f8f3e88fd53b795b110654a7c 100644
--- a/Embedding.Keras2/03-Prediction.ipynb
+++ b/Embedding.Keras3/03-Prediction.ipynb
@@ -6,8 +6,8 @@
"source": [
"<img width=\"800px\" src=\"../fidle/img/header.svg\"></img>\n",
"\n",
- "# <!-- TITLE --> [K2IMDB3] - Reload and reuse a saved model\n",
- "<!-- DESC --> Retrieving a saved model to perform a sentiment analysis (movie review), using Keras 2 and Tensorflow (obsolete)\n",
+ "# <!-- TITLE --> [K3IMDB3] - Reload and reuse a saved model\n",
+ "<!-- DESC --> Retrieving a saved model to perform a sentiment analysis (movie review), using Keras 3 and PyTorch\n",
"<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
"\n",
"## Objectives :\n",
@@ -34,24 +34,18 @@
"metadata": {},
"outputs": [],
"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",
+ "import os\n",
+ "os.environ['KERAS_BACKEND'] = 'torch'\n",
"\n",
- "import os,sys,h5py,json,re\n",
+ "import keras\n",
"\n",
- "from importlib import reload\n",
+ "import json,re\n",
+ "import numpy as np\n",
"\n",
"import fidle\n",
"\n",
"# Init Fidle environment\n",
- "run_id, run_dir, datasets_dir = fidle.init('K2IMDB3')"
+ "run_id, run_dir, datasets_dir = fidle.init('K3IMDB3')"
]
},
{
@@ -75,7 +69,7 @@
"vocab_size = 10000\n",
"review_len = 256\n",
"\n",
- "saved_models = './run/IMDB2'\n",
+ "saved_models = './run/K3IMDB2'\n",
"dictionaries_dir = './data'"
]
},
@@ -131,10 +125,8 @@
"source": [
"with open(f'{dictionaries_dir}/word_index.json', 'r') as fp:\n",
" word_index = json.load(fp)\n",
- " word_index = { w:int(i) for w,i in word_index.items() }\n",
- " print('Loaded. ', len(word_index), 'entries in word_index' )\n",
" index_word = { i:w for w,i in word_index.items() }\n",
- " print('Loaded. ', len(index_word), 'entries in index_word' )"
+ " print('Dictionaries loaded. ', len(word_index), 'entries' )"
]
},
{
@@ -152,6 +144,10 @@
"metadata": {},
"outputs": [],
"source": [
+ "start_char = 1 # Start of a sequence (padding is 0)\n",
+ "oov_char = 2 # Out-of-vocabulary\n",
+ "index_from = 3 # First word id\n",
+ "\n",
"nb_reviews = len(reviews)\n",
"x_data = []\n",
"\n",
@@ -159,17 +155,17 @@
"for review in reviews:\n",
" print('Words are : ', end='')\n",
" # ---- First index must be <start>\n",
- " index_review=[1]\n",
- " print('1 ', end='')\n",
+ " index_review=[start_char]\n",
+ " print(f'{start_char} ', end='')\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",
+ " # ---- Get the index - must be inside dict or is out of vocab (oov)\n",
+ " w_index = word_index.get(w, oov_char)\n",
+ " if w_index>vocab_size : w_index=oov_char\n",
" # ---- Add the index if < vocab_size\n",
" index_review.append(w_index)\n",
" print(f'{w_index} ', end='')\n",
@@ -199,9 +195,9 @@
"\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]), '(...)')"
+ " print(f'\\nText review {i} :', reviews[i])\n",
+ " print(f'tokens vector :', list(x_data[i][:imax]), '(...)')\n",
+ " print('Translation :', translate(x_data[i][:imax]), '(...)')"
]
},
{
@@ -217,7 +213,7 @@
"metadata": {},
"outputs": [],
"source": [
- "model = keras.models.load_model(f'{saved_models}/models/best_model.h5')"
+ "model = keras.models.load_model(f'{saved_models}/models/best_model.keras')"
]
},
{
diff --git a/Embedding.Keras2/04-Show-vectors.ipynb b/Embedding.Keras3/04-Show-vectors.ipynb
similarity index 87%
rename from Embedding.Keras2/04-Show-vectors.ipynb
rename to Embedding.Keras3/04-Show-vectors.ipynb
index 3ccd03089d85ba849730ff41d341f51cdbdf0ccb..862242e73586938969830771feb99ff824b8f6b3 100644
--- a/Embedding.Keras2/04-Show-vectors.ipynb
+++ b/Embedding.Keras3/04-Show-vectors.ipynb
@@ -6,8 +6,8 @@
"source": [
"<img width=\"800px\" src=\"../fidle/img/header.svg\"></img>\n",
"\n",
- "# <!-- TITLE --> [K2IMDB4] - Reload embedded vectors\n",
- "<!-- DESC --> Retrieving embedded vectors from our trained model, using Keras 2 and Tensorflow (obsolete)\n",
+ "# <!-- TITLE --> [K3IMDB4] - Reload embedded vectors\n",
+ "<!-- DESC --> Retrieving embedded vectors from our trained model, using Keras 3 and PyTorch\n",
"<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
"\n",
"## Objectives :\n",
@@ -33,24 +33,18 @@
"metadata": {},
"outputs": [],
"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",
+ "import os\n",
+ "os.environ['KERAS_BACKEND'] = 'torch'\n",
"\n",
- "import matplotlib.pyplot as plt\n",
- "import matplotlib\n",
- "import pandas as pd\n",
+ "import keras\n",
"\n",
- "import os,sys,h5py,json,re\n",
- "\n",
- "from importlib import reload\n",
+ "import json,re\n",
+ "import numpy as np\n",
"\n",
"import fidle\n",
"\n",
"# Init Fidle environment\n",
- "run_id, run_dir, datasets_dir = fidle.init('K2IMDB4')"
+ "run_id, run_dir, datasets_dir = fidle.init('K3IMDB4')"
]
},
{
@@ -74,7 +68,7 @@
"vocab_size = 5000\n",
"review_len = 256\n",
"\n",
- "saved_models = './run/IMDB2'\n",
+ "saved_models = './run/K3IMDB2'\n",
"dictionaries_dir = './data'"
]
},
@@ -114,14 +108,13 @@
"metadata": {},
"outputs": [],
"source": [
- "model = keras.models.load_model(f'{saved_models}/models/best_model.h5')\n",
+ "model = keras.models.load_model(f'{saved_models}/models/best_model.keras')\n",
"print('Model loaded.')\n",
"\n",
- "with open(f'{dictionaries_dir}/index_word.json', 'r') as fp:\n",
- " index_word = json.load(fp)\n",
- " index_word = { int(i):w for i,w in index_word.items() }\n",
- " word_index = { w:int(i) for i,w in index_word.items() }\n",
- " print('Dictionary loaded.')"
+ "with open(f'{dictionaries_dir}/word_index.json', 'r') as fp:\n",
+ " word_index = json.load(fp)\n",
+ " index_word = { i:w for w,i in word_index.items() }\n",
+ " print('Dictionaries loaded. ', len(word_index), 'entries' )"
]
},
{
diff --git a/Embedding.Keras2/05-LSTM-Keras.ipynb b/Embedding.Keras3/05-LSTM-Keras.ipynb
similarity index 78%
rename from Embedding.Keras2/05-LSTM-Keras.ipynb
rename to Embedding.Keras3/05-LSTM-Keras.ipynb
index 0b5fbef7e2aae25475a529b7e6a9cffc09892101..620c3c86335fd89a1a3a76669f4cdd7dea63b586 100644
--- a/Embedding.Keras2/05-LSTM-Keras.ipynb
+++ b/Embedding.Keras3/05-LSTM-Keras.ipynb
@@ -6,8 +6,8 @@
"source": [
"<img width=\"800px\" src=\"../fidle/img/header.svg\"></img>\n",
"\n",
- "# <!-- TITLE --> [K2IMDB5] - Sentiment analysis with a RNN network\n",
- "<!-- DESC --> Still the same problem, but with a network combining embedding and RNN, using Keras 2 and Tensorflow (obsolete)\n",
+ "# <!-- TITLE --> [K3IMDB5] - Sentiment analysis with a RNN network\n",
+ "<!-- DESC --> Still the same problem, but with a network combining embedding and RNN, using Keras 3 and PyTorch\n",
"<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
"\n",
"## Objectives :\n",
@@ -40,22 +40,19 @@
"metadata": {},
"outputs": [],
"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",
+ "import os\n",
+ "os.environ['KERAS_BACKEND'] = 'torch'\n",
"\n",
- "import matplotlib.pyplot as plt\n",
- "import matplotlib\n",
+ "import keras\n",
+ "import keras.datasets.imdb as imdb\n",
"\n",
- "import os,sys,h5py,json\n",
- "from importlib import reload\n",
+ "import json,re\n",
+ "import numpy as np\n",
"\n",
"import fidle\n",
"\n",
"# Init Fidle environment\n",
- "run_id, run_dir, datasets_dir = fidle.init('K2IMDB5')"
+ "run_id, run_dir, datasets_dir = fidle.init('K3IMDB5')"
]
},
{
@@ -135,10 +132,17 @@
"metadata": {},
"outputs": [],
"source": [
- "(x_train, y_train), (x_test, y_test) = imdb.load_data( num_words=vocab_size, skip_top=hide_most_frequently, seed= 42,)\n",
+ "# ----- Retrieve x,y\n",
+ "#\n",
+ "start_char = 1 # Start of a sequence (padding is 0)\n",
+ "oov_char = 2 # Out-of-vocabulary\n",
+ "index_from = 3 # First word id\n",
"\n",
- "y_train = np.asarray(y_train).astype('float32')\n",
- "y_test = np.asarray(y_test ).astype('float32')\n",
+ "(x_train, y_train), (x_test, y_test) = imdb.load_data( num_words = vocab_size, \n",
+ " skip_top = hide_most_frequently,\n",
+ " start_char = start_char, \n",
+ " oov_char = oov_char, \n",
+ " index_from = index_from)\n",
"\n",
"# ---- Rescale\n",
"#\n",
@@ -151,28 +155,8 @@
"#\n",
"print(\"Max(x_train,x_test) : \", fidle.utils.rmax([x_train,x_test]) )\n",
"print(\"Min(x_train,x_test) : \", fidle.utils.rmin([x_train,x_test]) )\n",
- "print(\"x_train : {} y_train : {}\".format(x_train.shape, y_train.shape))\n",
- "print(\"x_test : {} y_test : {}\".format(x_test.shape, y_test.shape))"
- ]
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "**About this dataset :**"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": [
- "print(\" Max(x_train,x_test) : \", fidle.utils.rmax([x_train,x_test]) )\n",
- "print(\" x_train : {} y_train : {}\".format(x_train.shape, y_train.shape))\n",
- "print(\" x_test : {} y_test : {}\".format(x_test.shape, y_test.shape))\n",
- "\n",
- "print('\\nReview example (x_train[12]) :\\n\\n',x_train[12])"
+ "print(\"Len(x_train) : \", len(x_train))\n",
+ "print(\"Len(x_test) : \", len(x_test))"
]
},
{
@@ -193,22 +177,16 @@
"outputs": [],
"source": [
"# ---- Retrieve dictionary {word:index}, and encode it in ascii\n",
+ "# Shift the dictionary from +3\n",
+ "# Add <pad>, <start> and <unknown> tags\n",
+ "# Create a reverse dictionary : {index:word}\n",
"#\n",
"word_index = imdb.get_word_index()\n",
- "\n",
- "# ---- Shift the dictionary from +3\n",
- "#\n",
- "word_index = {w:(i+3) for w,i in word_index.items()}\n",
- "\n",
- "# ---- Add <pad>, <start> and unknown tags\n",
- "#\n",
- "word_index.update( {'<pad>':0, '<start>':1, '<unknown>':2} )\n",
- "\n",
- "# ---- Create a reverse dictionary : {index:word}\n",
- "#\n",
+ "word_index = {w:(i+index_from) for w,i in word_index.items()}\n",
+ "word_index.update( {'<pad>':0, '<start>':1, '<unknown>':2, '<undef>':3,} )\n",
"index_word = {index:word for word,index in word_index.items()} \n",
"\n",
- "# ---- Add a nice function to transpose :\n",
+ "# ---- A nice function to transpose :\n",
"#\n",
"def dataset2text(review):\n",
" return ' '.join([index_word.get(i, '?') for i in review])"
@@ -284,34 +262,14 @@
"metadata": {},
"outputs": [],
"source": [
- "def get_model(dense_vector_size=128):\n",
- " \n",
- " model = keras.Sequential()\n",
- " model.add(keras.layers.Embedding(input_dim = vocab_size, output_dim = dense_vector_size))\n",
- " model.add(keras.layers.GRU(50))\n",
- " model.add(keras.layers.Dense(1, activation='sigmoid'))\n",
- "\n",
- " model.compile(optimizer = 'rmsprop',\n",
- " loss = 'binary_crossentropy',\n",
- " metrics = ['accuracy'])\n",
- " return model"
- ]
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "## Step 6 - Train the model\n",
- "### 6.1 - Get it"
- ]
- },
- {
- "cell_type": "code",
- "execution_count": null,
- "metadata": {},
- "outputs": [],
- "source": [
- "model = get_model(32)\n",
+ "model = keras.Sequential()\n",
+ "model.add(keras.layers.Embedding(input_dim = vocab_size, output_dim = dense_vector_size))\n",
+ "model.add(keras.layers.GRU(50))\n",
+ "model.add(keras.layers.Dense(1, activation='sigmoid'))\n",
+ "\n",
+ "model.compile(optimizer = 'rmsprop',\n",
+ " loss = 'binary_crossentropy',\n",
+ " metrics = ['accuracy'])\n",
"\n",
"model.summary()"
]
@@ -320,7 +278,8 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 6.2 - Add callback"
+ "## Step 6 - Train the model\n",
+ "### 6.1 - Add Callbacks"
]
},
{
@@ -330,16 +289,17 @@
"outputs": [],
"source": [
"os.makedirs(f'{run_dir}/models', mode=0o750, exist_ok=True)\n",
- "save_dir = f'{run_dir}/models/best_model.h5'\n",
- "savemodel_callback = tf.keras.callbacks.ModelCheckpoint(filepath=save_dir, verbose=0, save_best_only=True)"
+ "save_dir = f'{run_dir}/models/best_model.keras'\n",
+ "\n",
+ "savemodel_callback = keras.callbacks.ModelCheckpoint( filepath=save_dir, monitor='val_accuracy', mode='max', save_best_only=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 6.3 - Train it\n",
- "CPU : batch_size=128, epochs=10 : Need 9'30 (CPU, laptop)"
+ "### 6.2 - Train it\n",
+ "Note : With a scale=0.2, batch_size=128, epochs=10 => Need 4' on a cpu laptop"
]
},
{
@@ -348,18 +308,13 @@
"metadata": {},
"outputs": [],
"source": [
- "chrono = fidle.Chrono()\n",
- "chrono.start()\n",
- "\n",
"history = model.fit(x_train,\n",
" y_train,\n",
" epochs = epochs,\n",
" batch_size = batch_size,\n",
" validation_data = (x_test, y_test),\n",
" verbose = fit_verbosity,\n",
- " callbacks = [savemodel_callback])\n",
- "\n",
- "chrono.show()"
+ " callbacks = [savemodel_callback])"
]
},
{
@@ -392,7 +347,7 @@
"metadata": {},
"outputs": [],
"source": [
- "model = keras.models.load_model(f'{run_dir}/models/best_model.h5')\n",
+ "model = keras.models.load_model(f'{run_dir}/models/best_model.keras')\n",
"\n",
"# ---- Evaluate\n",
"score = model.evaluate(x_test, y_test, verbose=0)\n",
diff --git a/GTSRB.Keras3/04-Keras-cv.ipynb b/GTSRB.Keras3/04-Keras-cv.ipynb
index ad3e9e0047188b820ebceb28729a92c354fa43d6..6c88dc768d1356d4c945d2b18e8516d5519990e4 100644
--- a/GTSRB.Keras3/04-Keras-cv.ipynb
+++ b/GTSRB.Keras3/04-Keras-cv.ipynb
@@ -6,8 +6,8 @@
"source": [
"<img width=\"800px\" src=\"../fidle/img/header.svg\"></img>\n",
"\n",
- "# <!-- TITLE --> [K3GTSRB4] - Hight level example\n",
- "<!-- DESC --> Episode 4 : An example of using a pre-trained model (Keras-cv)\n",
+ "# <!-- TITLE --> [K3GTSRB4] - Hight level example (Keras-cv)\n",
+ "<!-- DESC --> An example of using a pre-trained model with Keras-cv\n",
"<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
"\n",
"## Objectives :\n",
@@ -22,7 +22,8 @@
"\n",
"## Step 1 - Import and init\n",
"\n",
- "The python environment required for this notebook is :\n",
+ "**ATTENTION :** A specific environment is required for this example (Which may require 6 GB). \n",
+ "This python environment required for this notebook is :\n",
"```\n",
"python3 -m venv fidle-kcv\n",
"pip install --upgrade keras-cv tensorflow torch torchvision torchaudio Matplotlib Jupyterlab\n",
diff --git a/MNIST.Keras3/02-CNN-MNIST.ipynb b/MNIST.Keras3/02-CNN-MNIST.ipynb
index ae6a3d0283fe06b236ca2c4a0fbb0a5f4697a65e..19c6b220d842a795297699eccda77e4373495a51 100644
--- a/MNIST.Keras3/02-CNN-MNIST.ipynb
+++ b/MNIST.Keras3/02-CNN-MNIST.ipynb
@@ -55,7 +55,7 @@
"# Init Fidle environment\n",
"import fidle\n",
"\n",
- "run_id, run_dir, datasets_dir = fidle.init('K3MNIST1')"
+ "run_id, run_dir, datasets_dir = fidle.init('K3MNIST2')"
]
},
{
diff --git a/README.ipynb b/README.ipynb
index 3b01dedb91291280fccca33c03bbd782e2899c50..7f16f1a9afd93eccf0d8e3e5edfb884f8ffa6beb 100644
--- a/README.ipynb
+++ b/README.ipynb
@@ -3,13 +3,13 @@
{
"cell_type": "code",
"execution_count": 1,
- "id": "9fccea2b",
+ "id": "1f828036",
"metadata": {
"execution": {
- "iopub.execute_input": "2024-01-16T21:04:27.072205Z",
- "iopub.status.busy": "2024-01-16T21:04:27.071380Z",
- "iopub.status.idle": "2024-01-16T21:04:27.081582Z",
- "shell.execute_reply": "2024-01-16T21:04:27.080749Z"
+ "iopub.execute_input": "2024-01-21T16:21:09.860108Z",
+ "iopub.status.busy": "2024-01-21T16:21:09.859792Z",
+ "iopub.status.idle": "2024-01-21T16:21:09.870962Z",
+ "shell.execute_reply": "2024-01-21T16:21:09.870075Z"
},
"jupyter": {
"source_hidden": true
@@ -52,7 +52,7 @@
"For more information, you can contact us at : \n",
"[<img width=\"200px\" style=\"vertical-align:middle\" src=\"fidle/img/00-Mail_contact.svg\"></img>](#top)\n",
"\n",
- "Current Version : <!-- VERSION_BEGIN -->2.5.1<!-- VERSION_END -->\n",
+ "Current Version : <!-- VERSION_BEGIN -->2.5.4<!-- VERSION_END -->\n",
"\n",
"\n",
"## Course materials\n",
@@ -73,7 +73,7 @@
"The world of Deep Learning is changing very fast !\n",
"\n",
"<!-- TOC_BEGIN -->\n",
- "<!-- Automatically generated on : 16/01/24 22:04:25 -->\n",
+ "<!-- Automatically generated on : 21/01/24 17:21:08 -->\n",
"\n",
"### Linear and logistic regression\n",
"- **[LINR1](LinearReg/01-Linear-Regression.ipynb)** - [Linear regression with direct resolution](LinearReg/01-Linear-Regression.ipynb) \n",
@@ -117,7 +117,7 @@
"- **[PMNIST1](MNIST.PyTorch/01-DNN-MNIST_PyTorch.ipynb)** - [Simple classification with DNN](MNIST.PyTorch/01-DNN-MNIST_PyTorch.ipynb) \n",
"Example of classification with a fully connected neural network, using Pytorch\n",
"\n",
- "### MNIST classification (DNN,CNN), using Lightning\n",
+ "### MNIST classification (DNN,CNN), using PyTorch/Lightning\n",
"- **[LMNIST2](MNIST.Lightning/01-DNN-MNIST_Lightning.ipynb)** - [Simple classification with DNN](MNIST.Lightning/01-DNN-MNIST_Lightning.ipynb) \n",
"An example of classification using a dense neural network for the famous MNIST dataset, using PyTorch Lightning\n",
"- **[LMNIST2](MNIST.Lightning/02-CNN-MNIST_Lightning.ipynb)** - [Simple classification with CNN](MNIST.Lightning/02-CNN-MNIST_Lightning.ipynb) \n",
@@ -130,28 +130,28 @@
"Episode 2 : First convolutions and first classification of our traffic signs, using Keras3\n",
"- **[K3GTSRB3](GTSRB.Keras3/03-Better-convolutions.ipynb)** - [Training monitoring](GTSRB.Keras3/03-Better-convolutions.ipynb) \n",
"Episode 3 : Monitoring, analysis and check points during a training session, using Keras3\n",
- "- **[K3GTSRB4](GTSRB.Keras3/04-Keras-cv.ipynb)** - [Hight level example](GTSRB.Keras3/04-Keras-cv.ipynb) \n",
- "Episode 4 : An example of using a pre-trained model (Keras-cv)\n",
+ "- **[K3GTSRB4](GTSRB.Keras3/04-Keras-cv.ipynb)** - [Hight level example (Keras-cv)](GTSRB.Keras3/04-Keras-cv.ipynb) \n",
+ "An example of using a pre-trained model with Keras-cv\n",
"- **[K3GTSRB10](GTSRB.Keras3/batch_oar.sh)** - [OAR batch script submission](GTSRB.Keras3/batch_oar.sh) \n",
"Bash script for an OAR batch submission of an ipython code\n",
"- **[K3GTSRB11](GTSRB.Keras3/batch_slurm.sh)** - [SLURM batch script](GTSRB.Keras3/batch_slurm.sh) \n",
"Bash script for a Slurm batch submission of an ipython code\n",
"\n",
- "### Sentiment analysis with word embedding, using Keras2 (obsolete)\n",
- "- **[K2IMDB1](Embedding.Keras2/01-One-hot-encoding.ipynb)** - [Sentiment analysis with hot-one encoding](Embedding.Keras2/01-One-hot-encoding.ipynb) \n",
- "A basic example of sentiment analysis with sparse encoding, using a dataset from Internet Movie Database (IMDB), using Keras 2 and Tensorflow (obsolete)\n",
- "- **[K2IMDB2](Embedding.Keras2/02-Keras-embedding.ipynb)** - [Sentiment analysis with text embedding](Embedding.Keras2/02-Keras-embedding.ipynb) \n",
- "A very classical example of word embedding with a dataset from Internet Movie Database (IMDB), using Keras 2 and Tensorflow (obsolete)\n",
- "- **[K2IMDB3](Embedding.Keras2/03-Prediction.ipynb)** - [Reload and reuse a saved model](Embedding.Keras2/03-Prediction.ipynb) \n",
- "Retrieving a saved model to perform a sentiment analysis (movie review), using Keras 2 and Tensorflow (obsolete)\n",
- "- **[K2IMDB4](Embedding.Keras2/04-Show-vectors.ipynb)** - [Reload embedded vectors](Embedding.Keras2/04-Show-vectors.ipynb) \n",
- "Retrieving embedded vectors from our trained model, using Keras 2 and Tensorflow (obsolete)\n",
- "- **[K2IMDB5](Embedding.Keras2/05-LSTM-Keras.ipynb)** - [Sentiment analysis with a RNN network](Embedding.Keras2/05-LSTM-Keras.ipynb) \n",
- "Still the same problem, but with a network combining embedding and RNN, using Keras 2 and Tensorflow (obsolete)\n",
- "\n",
- "### Time series with Recurrent Neural Network (RNN), using Keras2 (obsolete)\n",
- "- **[K2LADYB1](RNN.Keras2/LADYB1-Ladybug.ipynb)** - [Prediction of a 2D trajectory via RNN](RNN.Keras2/LADYB1-Ladybug.ipynb) \n",
- "Artificial dataset generation and prediction attempt via a recurrent network, using Keras 2 and Tensorflow (obsolete)\n",
+ "### Sentiment analysis with word embedding, using Keras3/PyTorch\n",
+ "- **[K3IMDB1](Embedding.Keras3/01-One-hot-encoding.ipynb)** - [Sentiment analysis with hot-one encoding](Embedding.Keras3/01-One-hot-encoding.ipynb) \n",
+ "A basic example of sentiment analysis with sparse encoding, using a dataset from Internet Movie Database (IMDB), using Keras 3 on PyTorch\n",
+ "- **[K3IMDB2](Embedding.Keras3/02-Keras-embedding.ipynb)** - [Sentiment analysis with text embedding](Embedding.Keras3/02-Keras-embedding.ipynb) \n",
+ "A very classical example of word embedding with a dataset from Internet Movie Database (IMDB), using Keras 3 on PyTorch\n",
+ "- **[K3IMDB3](Embedding.Keras3/03-Prediction.ipynb)** - [Reload and reuse a saved model](Embedding.Keras3/03-Prediction.ipynb) \n",
+ "Retrieving a saved model to perform a sentiment analysis (movie review), using Keras 3 and PyTorch\n",
+ "- **[K3IMDB4](Embedding.Keras3/04-Show-vectors.ipynb)** - [Reload embedded vectors](Embedding.Keras3/04-Show-vectors.ipynb) \n",
+ "Retrieving embedded vectors from our trained model, using Keras 3 and PyTorch\n",
+ "- **[K3IMDB5](Embedding.Keras3/05-LSTM-Keras.ipynb)** - [Sentiment analysis with a RNN network](Embedding.Keras3/05-LSTM-Keras.ipynb) \n",
+ "Still the same problem, but with a network combining embedding and RNN, using Keras 3 and PyTorch\n",
+ "\n",
+ "### Time series with Recurrent Neural Network (RNN), using Keras3/PyTorch\n",
+ "- **[K3LADYB1](RNN.Keras3/01-Ladybug.ipynb)** - [Prediction of a 2D trajectory via RNN](RNN.Keras3/01-Ladybug.ipynb) \n",
+ "Artificial dataset generation and prediction attempt via a recurrent network, using Keras 3 and PyTorch\n",
"\n",
"### Sentiment analysis with transformer, using PyTorch\n",
"- **[TRANS1](Transformers.PyTorch/01-Distilbert.ipynb)** - [IMDB, Sentiment analysis with Transformers ](Transformers.PyTorch/01-Distilbert.ipynb) \n",
@@ -256,7 +256,7 @@
"from IPython.display import display,Markdown\n",
"display(Markdown(open('README.md', 'r').read()))\n",
"#\n",
- "# This README is visible under Jupiter Lab ;-)# Automatically generated on : 16/01/24 22:04:26"
+ "# This README is visible under Jupiter Lab ;-)# Automatically generated on : 21/01/24 17:21:08"
]
}
],
diff --git a/README.md b/README.md
index 5393036c2568974352d402c3189935c60562818e..9295f816a0521507969df52c9b933e9f919f58c8 100644
--- a/README.md
+++ b/README.md
@@ -31,7 +31,7 @@ For more information, see **https://fidle.cnrs.fr** :
For more information, you can contact us at :
[<img width="200px" style="vertical-align:middle" src="fidle/img/00-Mail_contact.svg"></img>](#top)
-Current Version : <!-- VERSION_BEGIN -->2.5.1<!-- VERSION_END -->
+Current Version : <!-- VERSION_BEGIN -->2.5.4<!-- VERSION_END -->
## Course materials
@@ -52,7 +52,7 @@ For these reason, they are kept as examples, while we develop the Keras3/PyTorch
The world of Deep Learning is changing very fast !
<!-- TOC_BEGIN -->
-<!-- Automatically generated on : 16/01/24 22:04:25 -->
+<!-- Automatically generated on : 21/01/24 17:21:08 -->
### Linear and logistic regression
- **[LINR1](LinearReg/01-Linear-Regression.ipynb)** - [Linear regression with direct resolution](LinearReg/01-Linear-Regression.ipynb)
@@ -96,7 +96,7 @@ An example of classification using a convolutional neural network for the famous
- **[PMNIST1](MNIST.PyTorch/01-DNN-MNIST_PyTorch.ipynb)** - [Simple classification with DNN](MNIST.PyTorch/01-DNN-MNIST_PyTorch.ipynb)
Example of classification with a fully connected neural network, using Pytorch
-### MNIST classification (DNN,CNN), using Lightning
+### MNIST classification (DNN,CNN), using PyTorch/Lightning
- **[LMNIST2](MNIST.Lightning/01-DNN-MNIST_Lightning.ipynb)** - [Simple classification with DNN](MNIST.Lightning/01-DNN-MNIST_Lightning.ipynb)
An example of classification using a dense neural network for the famous MNIST dataset, using PyTorch Lightning
- **[LMNIST2](MNIST.Lightning/02-CNN-MNIST_Lightning.ipynb)** - [Simple classification with CNN](MNIST.Lightning/02-CNN-MNIST_Lightning.ipynb)
@@ -109,28 +109,28 @@ Episode 1 : Analysis of the GTSRB dataset and creation of an enhanced dataset
Episode 2 : First convolutions and first classification of our traffic signs, using Keras3
- **[K3GTSRB3](GTSRB.Keras3/03-Better-convolutions.ipynb)** - [Training monitoring](GTSRB.Keras3/03-Better-convolutions.ipynb)
Episode 3 : Monitoring, analysis and check points during a training session, using Keras3
-- **[K3GTSRB4](GTSRB.Keras3/04-Keras-cv.ipynb)** - [Hight level example](GTSRB.Keras3/04-Keras-cv.ipynb)
-Episode 4 : An example of using a pre-trained model (Keras-cv)
+- **[K3GTSRB4](GTSRB.Keras3/04-Keras-cv.ipynb)** - [Hight level example (Keras-cv)](GTSRB.Keras3/04-Keras-cv.ipynb)
+An example of using a pre-trained model with Keras-cv
- **[K3GTSRB10](GTSRB.Keras3/batch_oar.sh)** - [OAR batch script submission](GTSRB.Keras3/batch_oar.sh)
Bash script for an OAR batch submission of an ipython code
- **[K3GTSRB11](GTSRB.Keras3/batch_slurm.sh)** - [SLURM batch script](GTSRB.Keras3/batch_slurm.sh)
Bash script for a Slurm batch submission of an ipython code
-### Sentiment analysis with word embedding, using Keras2 (obsolete)
-- **[K2IMDB1](Embedding.Keras2/01-One-hot-encoding.ipynb)** - [Sentiment analysis with hot-one encoding](Embedding.Keras2/01-One-hot-encoding.ipynb)
-A basic example of sentiment analysis with sparse encoding, using a dataset from Internet Movie Database (IMDB), using Keras 2 and Tensorflow (obsolete)
-- **[K2IMDB2](Embedding.Keras2/02-Keras-embedding.ipynb)** - [Sentiment analysis with text embedding](Embedding.Keras2/02-Keras-embedding.ipynb)
-A very classical example of word embedding with a dataset from Internet Movie Database (IMDB), using Keras 2 and Tensorflow (obsolete)
-- **[K2IMDB3](Embedding.Keras2/03-Prediction.ipynb)** - [Reload and reuse a saved model](Embedding.Keras2/03-Prediction.ipynb)
-Retrieving a saved model to perform a sentiment analysis (movie review), using Keras 2 and Tensorflow (obsolete)
-- **[K2IMDB4](Embedding.Keras2/04-Show-vectors.ipynb)** - [Reload embedded vectors](Embedding.Keras2/04-Show-vectors.ipynb)
-Retrieving embedded vectors from our trained model, using Keras 2 and Tensorflow (obsolete)
-- **[K2IMDB5](Embedding.Keras2/05-LSTM-Keras.ipynb)** - [Sentiment analysis with a RNN network](Embedding.Keras2/05-LSTM-Keras.ipynb)
-Still the same problem, but with a network combining embedding and RNN, using Keras 2 and Tensorflow (obsolete)
-
-### Time series with Recurrent Neural Network (RNN), using Keras2 (obsolete)
-- **[K2LADYB1](RNN.Keras2/LADYB1-Ladybug.ipynb)** - [Prediction of a 2D trajectory via RNN](RNN.Keras2/LADYB1-Ladybug.ipynb)
-Artificial dataset generation and prediction attempt via a recurrent network, using Keras 2 and Tensorflow (obsolete)
+### Sentiment analysis with word embedding, using Keras3/PyTorch
+- **[K3IMDB1](Embedding.Keras3/01-One-hot-encoding.ipynb)** - [Sentiment analysis with hot-one encoding](Embedding.Keras3/01-One-hot-encoding.ipynb)
+A basic example of sentiment analysis with sparse encoding, using a dataset from Internet Movie Database (IMDB), using Keras 3 on PyTorch
+- **[K3IMDB2](Embedding.Keras3/02-Keras-embedding.ipynb)** - [Sentiment analysis with text embedding](Embedding.Keras3/02-Keras-embedding.ipynb)
+A very classical example of word embedding with a dataset from Internet Movie Database (IMDB), using Keras 3 on PyTorch
+- **[K3IMDB3](Embedding.Keras3/03-Prediction.ipynb)** - [Reload and reuse a saved model](Embedding.Keras3/03-Prediction.ipynb)
+Retrieving a saved model to perform a sentiment analysis (movie review), using Keras 3 and PyTorch
+- **[K3IMDB4](Embedding.Keras3/04-Show-vectors.ipynb)** - [Reload embedded vectors](Embedding.Keras3/04-Show-vectors.ipynb)
+Retrieving embedded vectors from our trained model, using Keras 3 and PyTorch
+- **[K3IMDB5](Embedding.Keras3/05-LSTM-Keras.ipynb)** - [Sentiment analysis with a RNN network](Embedding.Keras3/05-LSTM-Keras.ipynb)
+Still the same problem, but with a network combining embedding and RNN, using Keras 3 and PyTorch
+
+### Time series with Recurrent Neural Network (RNN), using Keras3/PyTorch
+- **[K3LADYB1](RNN.Keras3/01-Ladybug.ipynb)** - [Prediction of a 2D trajectory via RNN](RNN.Keras3/01-Ladybug.ipynb)
+Artificial dataset generation and prediction attempt via a recurrent network, using Keras 3 and PyTorch
### Sentiment analysis with transformer, using PyTorch
- **[TRANS1](Transformers.PyTorch/01-Distilbert.ipynb)** - [IMDB, Sentiment analysis with Transformers ](Transformers.PyTorch/01-Distilbert.ipynb)
diff --git a/RNN.Keras2/LADYB1-Ladybug.ipynb b/RNN.Keras3/01-Ladybug.ipynb
similarity index 82%
rename from RNN.Keras2/LADYB1-Ladybug.ipynb
rename to RNN.Keras3/01-Ladybug.ipynb
index 806ae1b1f31fe98d0cc7d4361887fb08bbb39ab0..43ed94d101bd6f3ef8c3792c9086ca45da546d61 100644
--- a/RNN.Keras2/LADYB1-Ladybug.ipynb
+++ b/RNN.Keras3/01-Ladybug.ipynb
@@ -6,8 +6,8 @@
"source": [
"<img width=\"800px\" src=\"../fidle/img/header.svg\"></img>\n",
"\n",
- "# <!-- TITLE --> [K2LADYB1] - Prediction of a 2D trajectory via RNN\n",
- "<!-- DESC --> Artificial dataset generation and prediction attempt via a recurrent network, using Keras 2 and Tensorflow (obsolete)\n",
+ "# <!-- TITLE --> [K3LADYB1] - Prediction of a 2D trajectory via RNN\n",
+ "<!-- DESC --> Artificial dataset generation and prediction attempt via a recurrent network, using Keras 3 and PyTorch\n",
"<!-- AUTHOR : Jean-Luc Parouty (CNRS/SIMaP) -->\n",
"\n",
"## Objectives :\n",
@@ -17,7 +17,7 @@
"\n",
" - Generate an artificial dataset\n",
" - dataset preparation\n",
- " - Doing our training\n",
+ " - Doing our testing\n",
" - Making predictions\n",
"\n",
"## Step 1 - Import and init\n",
@@ -30,26 +30,18 @@
"metadata": {},
"outputs": [],
"source": [
- "import tensorflow as tf\n",
- "from tensorflow import keras\n",
- "from tensorflow.keras.callbacks import TensorBoard\n",
- "from tensorflow.keras.preprocessing.sequence import TimeseriesGenerator\n",
+ "import os\n",
+ "os.environ['KERAS_BACKEND'] = 'torch'\n",
"\n",
+ "import keras\n",
"import numpy as np\n",
- "import math, random\n",
- "from math import sin,cos,pi\n",
- "import matplotlib.pyplot as plt\n",
- "\n",
- "import pandas as pd\n",
- "import h5py, json\n",
- "import os,time,sys\n",
- "\n",
- "from importlib import reload\n",
+ "from math import cos, sin\n",
+ "import random\n",
"\n",
"import fidle\n",
"\n",
"# Init Fidle environment\n",
- "run_id, run_dir, datasets_dir = fidle.init('K2LADYB1')"
+ "run_id, run_dir, datasets_dir = fidle.init('K3LADYB1')"
]
},
{
@@ -68,7 +60,7 @@
"# ---- About dataset\n",
"#\n",
"max_t = 1000\n",
- "delta_t = 0.02\n",
+ "delta_t = 0.01\n",
"features_len = 2\n",
"\n",
"\n",
@@ -121,7 +113,7 @@
" ladybug_init.params_y = [ random.gauss(0.,1.) for u in range(8)]\n",
" \n",
"def ladybug_move(t):\n",
- " k=0.5\n",
+ "\n",
" [ax1, ax2, ax3, ax4, kx1, kx2, kx3, kx4] = ladybug_init.params_x\n",
" [ay1, ay2, ay3, ay4, ky1, ky2, ky3, ky4] = ladybug_init.params_y\n",
" \n",
@@ -209,7 +201,7 @@
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 2.4 - Prepare some nice data generator"
+ "### 2.4 - Prepare sequences from datasets"
]
},
{
@@ -218,32 +210,26 @@
"metadata": {},
"outputs": [],
"source": [
- "# ---- Train generator\n",
- "#\n",
- "train_generator = TimeseriesGenerator(x_train, x_train, length=sequence_len, batch_size=batch_size)\n",
- "test_generator = TimeseriesGenerator(x_test, x_test, length=sequence_len, batch_size=batch_size)\n",
- "\n",
- "# ---- About\n",
+ "# ---- Create sequences and labels for train and test\n",
"#\n",
- "fidle.utils.subtitle('About the splitting of our dataset :')\n",
+ "xs_train, ys_train=[],[]\n",
+ "all_i = np.random.permutation( len(x_train) - sequence_len - 1 )\n",
"\n",
- "x,y=train_generator[0]\n",
- "print(f'Number of batch trains available : ', len(train_generator))\n",
- "print('batch x shape : ',x.shape)\n",
- "print('batch y shape : ',y.shape)\n",
+ "for i in all_i:\n",
+ " xs_train.append( x_train[ i : i+sequence_len ] )\n",
+ " ys_train.append( x_train[ i+sequence_len+1 ] )\n",
+ " \n",
+ "xs_test, ys_test=[],[]\n",
+ "for i in range( len(x_test) - sequence_len - 1):\n",
+ " xs_test.append( x_test[ i : i+sequence_len ] )\n",
+ " ys_test.append( x_test[ i+sequence_len+1 ] )\n",
"\n",
- "x,y=train_generator[0]\n",
- "fidle.utils.subtitle('What a batch looks like (x) :')\n",
- "fidle.utils.np_print(x[0] )\n",
- "fidle.utils.subtitle('What a batch looks like (y) :')\n",
- "fidle.utils.np_print(y[0])"
- ]
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "## Step 3 - Create a model"
+ "# ---- Convert to numpy / float16\n",
+ " \n",
+ "xs_train = np.array(xs_train, dtype='float16')\n",
+ "ys_train = np.array(ys_train, dtype='float16')\n",
+ "xs_test = np.array(xs_test, dtype='float16')\n",
+ "ys_test = np.array(ys_test, dtype='float16')\n"
]
},
{
@@ -252,26 +238,23 @@
"metadata": {},
"outputs": [],
"source": [
- "model = keras.models.Sequential()\n",
- "model.add( keras.layers.InputLayer(input_shape=(sequence_len, features_len)) )\n",
- "model.add( keras.layers.GRU(200, return_sequences=False, activation='relu') )\n",
- "model.add( keras.layers.Dense(features_len) )\n",
+ "fidle.utils.subtitle('About the splitting of our dataset :')\n",
"\n",
- "model.summary()"
- ]
- },
- {
- "cell_type": "markdown",
- "metadata": {},
- "source": [
- "## Step 4 - Compile and run"
+ "print('Number of sequences : ', len(xs_train))\n",
+ "print('xs_train shape : ',xs_train.shape)\n",
+ "print('ys_train shape : ',ys_train.shape)\n",
+ "\n",
+ "fidle.utils.subtitle('What an xs look like :')\n",
+ "fidle.utils.np_print(xs_train[10] )\n",
+ "fidle.utils.subtitle('What an ys look like :')\n",
+ "fidle.utils.np_print(ys_train[10])"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 4.1 - Add callback"
+ "## Step 3 - Create a model"
]
},
{
@@ -280,17 +263,24 @@
"metadata": {},
"outputs": [],
"source": [
- "os.makedirs(f'{run_dir}/models', mode=0o750, exist_ok=True)\n",
- "save_path = f'{run_dir}/models/best_model.h5'\n",
+ "model = keras.models.Sequential()\n",
+ "model.add( keras.layers.InputLayer(shape=(sequence_len, features_len)) )\n",
+ "model.add( keras.layers.GRU(200, return_sequences=False, activation='relu') )\n",
+ "model.add( keras.layers.Dense(features_len) )\n",
+ "\n",
+ "model.summary()\n",
"\n",
- "bestmodel_callback = tf.keras.callbacks.ModelCheckpoint(filepath=save_path, verbose=0, save_best_only=True)"
+ "model.compile(optimizer='rmsprop', \n",
+ " loss='mse', \n",
+ " metrics = ['mae'] )"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 4.2 - Compile"
+ "## Step 4 - Train the model\n",
+ "### 4.1 Add Callbacks"
]
},
{
@@ -299,17 +289,18 @@
"metadata": {},
"outputs": [],
"source": [
- "model.compile(optimizer='rmsprop', \n",
- " loss='mse', \n",
- " metrics = ['mae'] )"
+ "os.makedirs(f'{run_dir}/models', mode=0o750, exist_ok=True)\n",
+ "save_dir = f'{run_dir}/models/best_model.keras'\n",
+ "\n",
+ "savemodel_callback = keras.callbacks.ModelCheckpoint( filepath=save_dir, monitor='val_mae', mode='max', save_best_only=True)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
- "### 4.3 - Fit\n",
- "3' with a CPU (laptop) "
+ "### 4.2 - Train it\n",
+ "Need 3' on a cpu laptop"
]
},
{
@@ -321,11 +312,11 @@
"chrono=fidle.Chrono()\n",
"chrono.start()\n",
"\n",
- "history=model.fit(train_generator, \n",
+ "history=model.fit(xs_train,ys_train,\n",
" epochs = epochs, \n",
" verbose = fit_verbosity,\n",
- " validation_data = test_generator,\n",
- " callbacks = [bestmodel_callback])\n",
+ " validation_data = (xs_test, ys_test),\n",
+ " callbacks = [savemodel_callback])\n",
"\n",
"chrono.show()"
]
@@ -359,7 +350,7 @@
"metadata": {},
"outputs": [],
"source": [
- "loaded_model = tf.keras.models.load_model(save_path)\n",
+ "loaded_model = keras.models.load_model(f'{run_dir}/models/best_model.keras')\n",
"print('Loaded.')"
]
},
diff --git a/fidle/about.yml b/fidle/about.yml
index ae56d2cb4b20b5551777093b06ba2af5e53ca19e..810344c14f0ca5913f817ac6631e76a8a2bf7feb 100644
--- a/fidle/about.yml
+++ b/fidle/about.yml
@@ -13,7 +13,7 @@
#
# This file describes the notebooks used by the Fidle training.
-version: 2.5.1
+version: 2.5.4
content: notebooks
name: Notebooks Fidle
description: All notebooks used by the Fidle training
@@ -32,10 +32,10 @@ toc:
Wine.Lightning: Wine Quality prediction (DNN), using PyTorch/Lightning
MNIST.Keras3: MNIST classification (DNN,CNN), using Keras3/PyTorch
MNIST.PyTorch: MNIST classification (DNN,CNN), using PyTorch
- MNIST.Lightning: MNIST classification (DNN,CNN), using Lightning
+ MNIST.Lightning: MNIST classification (DNN,CNN), using PyTorch/Lightning
GTSRB.Keras3: Images classification GTSRB with Convolutional Neural Networks (CNN), using Keras3/PyTorch
- Embedding.Keras2: Sentiment analysis with word embedding, using Keras2 (obsolete)
- RNN.Keras2: Time series with Recurrent Neural Network (RNN), using Keras2 (obsolete)
+ Embedding.Keras3: Sentiment analysis with word embedding, using Keras3/PyTorch
+ RNN.Keras3: Time series with Recurrent Neural Network (RNN), using Keras3/PyTorch
Transformers.PyTorch: Sentiment analysis with transformer, using PyTorch
AE.Keras2: Unsupervised learning with an autoencoder neural network (AE), using Keras2 (obsolete)
VAE.Keras2: Generative network with Variational Autoencoder (VAE), using Keras2 (obsolete)
diff --git a/fidle/ci/default.yml b/fidle/ci/default.yml
index 46dc911ae36ad16f3a160ae8adc8942d162da648..f8c8ee262b73969cd942d874ae066cc5751674e5 100644
--- a/fidle/ci/default.yml
+++ b/fidle/ci/default.yml
@@ -1,6 +1,6 @@
campain:
version: '1.0'
- description: Automatically generated ci profile (16/01/24 22:04:25)
+ description: Automatically generated ci profile (21/01/24 17:21:08)
directory: ./campains/default
existing_notebook: 'remove # remove|skip'
report_template: 'fidle # fidle|default'
@@ -116,18 +116,18 @@ K3GTSRB4:
notebook: GTSRB.Keras3/04-Keras-cv.ipynb
#
-# ------------ Embedding.Keras2
+# ------------ Embedding.Keras3
#
-K2IMDB1:
- notebook: Embedding.Keras2/01-One-hot-encoding.ipynb
+K3IMDB1:
+ notebook: Embedding.Keras3/01-One-hot-encoding.ipynb
overrides:
vocab_size: default
hide_most_frequently: default
batch_size: default
epochs: default
fit_verbosity: default
-K2IMDB2:
- notebook: Embedding.Keras2/02-Keras-embedding.ipynb
+K3IMDB2:
+ notebook: Embedding.Keras3/02-Keras-embedding.ipynb
overrides:
vocab_size: default
hide_most_frequently: default
@@ -137,22 +137,22 @@ K2IMDB2:
epochs: default
output_dir: default
fit_verbosity: default
-K2IMDB3:
- notebook: Embedding.Keras2/03-Prediction.ipynb
+K3IMDB3:
+ notebook: Embedding.Keras3/03-Prediction.ipynb
overrides:
vocab_size: default
review_len: default
saved_models: default
dictionaries_dir: default
-K2IMDB4:
- notebook: Embedding.Keras2/04-Show-vectors.ipynb
+K3IMDB4:
+ notebook: Embedding.Keras3/04-Show-vectors.ipynb
overrides:
vocab_size: default
review_len: default
saved_models: default
dictionaries_dir: default
-K2IMDB5:
- notebook: Embedding.Keras2/05-LSTM-Keras.ipynb
+K3IMDB5:
+ notebook: Embedding.Keras3/05-LSTM-Keras.ipynb
overrides:
vocab_size: default
hide_most_frequently: default
@@ -164,10 +164,10 @@ K2IMDB5:
scale: default
#
-# ------------ RNN.Keras2
+# ------------ RNN.Keras3
#
-K2LADYB1:
- notebook: RNN.Keras2/LADYB1-Ladybug.ipynb
+K3LADYB1:
+ notebook: RNN.Keras3/01-Ladybug.ipynb
overrides:
scale: default
train_prop: default