Add documentation for initializations and datasets

docs/sources/datasets.md

@@ -0,0 +1,78 @@
+# Datasets
+
+## CIFAR10 small image classification
+
+`keras.datasets.cifar10`
+
+Dataset of 50,000 32x32 color images, labeled over 10 categories.
+
+### Usage:
+
+```python
+(X_train, y_train), (X_test, y_test) = cifar10.load_data(test_split=0.1, seed=113)
+```
+
+__Returns:__
+
+- X_train, X_test: uint8 array of RGB image data with shape (nb_samples, 3, 32, 32).
+- y_train, y_test: uint8 array of category labels (integers in range 0-9) with shape (nb_samples,).
+
+__Arguments:__
+
+- test_split: float. Fraction of the dataset to be used as test data.
+- seed: int. Seed for reproducible data shuffling.
+
+## IMDB Movie reviews sentiment classification
+
+`keras.datasets.imdb`
+
+Dataset of 25,000 movies reviews from IMDB, labeled by sentiment (positive/negative). Reviews have been preprocessed, and each review is encoded as a sequence of word indexes (integers). For convenience, words are indexed by overall frequency in the dataset, so that for instance the integer "3" encodes the 3rd most frequent word in the data. This allows for quick filtering operations such as: "only consider the top 10,000 most common words, but eliminate the top 20 most common words".
+
+As a convention, "0" does not stand for a specific word, but instead is used to encode any unknown word.
+
+### Usage:
+
+```python
+(X_train, y_train), (X_test, y_test) = imdb.load_data(path="imdb.pkl", \
+nb_words=None, skip_top=0, maxlen=None, test_split=0.1, seed=113)`
+```
+__Returns:__
+
+- X_train, X_test: list of sequences, which are lists of indexes (integers). If the nb_words argument was specific, the maximum possible index value is nb_words-1. If the maxlen argument was specified, the largest possible sequence length is maxlen.
+- y_train, y_test: list of integer labels (1 or 0). 
+
+__Arguments:__
+
+- path: if you do have the data locally (at `'~/.keras/datasets/' + path`), if will be downloaded to this location (in cPickle format).
+- nb_words: integer or None. Top most frequent words to consider. Any less frequent word will appear as 0 in the sequence data.
+- skip_top: integer. Top most frequent words to ignore (they will appear as 0s in the sequence data).
+- maxlen: int. Maximum sequence length. Any longer sequence will be truncated.
+- test_split: float. Fraction of the dataset to be used as test data.
+- seed: int. Seed for reproducible data shuffling.
+
+## Reuters newswire topics classification
+
+`keras.datasets.reuters`
+
+Dataset of 11,228 newswires from Reuters, labeled over 46 topics. As with the IMDB dataset, each wire is encoded as a sequence of word indexes (same conventions).
+
+### Usage:
+
+```python
+(X_train, y_train), (X_test, y_test) = reuters.load_data(path="reuters.pkl", \
+nb_words=None, skip_top=0, maxlen=None, test_split=0.1, seed=113)`
+```
+
+The specifications are the same as that of the IMDB dataset.
+
+This dataset also makes available the word index used for encoding the sequences:
+
+```python
+word_index = reuters.get_word_index(path="reuters_word_index.pkl")
+```
+
+__Returns:__ A dictionary where key are words (str) and values are indexes (integer). eg. `word_index["giraffe"]` might return `1234`. 
+
+__Arguments:__
+
+- path: if you do have the index file locally (at `'~/.keras/datasets/' + path`), if will be downloaded to this location (in cPickle format).

docs/sources/initializations.md

@@ -0,0 +1,18 @@
+# Initializations
+
+## Usage of initializations
+
+Initializations define the probability distribution used to set the initial random weights of Keras layers.
+
+The keyword arguments used for passing initializations to layers will depend on the layer. Usually it is simply `init`:
+
+```python
+model.add(Dense(64, 64, init='uniform'))
+```
+
+## Available initializations
+
+- __uniform__
+- __normal__
+- __orthogonal__: use with square 2D layers (`shape[0] == shape[1]`).
+- __zero__