Updated / Deleted some examples

example/multivariate_time_series/README.md

@@ -3,6 +3,8 @@
 - This repo contains an MXNet implementation of [this](https://arxiv.org/pdf/1703.07015.pdf) state of the art time series forecasting model.
 - You can find my blog post on the model [here](https://opringle.github.io/2018/01/05/deep_learning_multivariate_ts.html)
 
+- A Gluon implementation is available [here](https://github.com/safrooze/LSTNet-Gluon)
+
 ![](./docs/model_architecture.png)
 
 ## Running the code
@@ -22,7 +24,7 @@
 
 ## Hyperparameters
 
-The default arguements in `lstnet.py` achieve equivolent performance to the published results. For other datasets, the following hyperparameters provide a good starting point:
+The default arguements in `lstnet.py` achieve equivalent performance to the published results. For other datasets, the following hyperparameters provide a good starting point:
 
 - q = {2^0, 2^1, ... , 2^9} (1 week is typical value)
 - Convolutional num filters  = {50, 100, 200}

example/named_entity_recognition/README.md

@@ -11,7 +11,6 @@ To reproduce the preprocessed training data:
 
 1. Download and unzip the data: https://www.kaggle.com/abhinavwalia95/entity-annotated-corpus/downloads/ner_dataset.csv
 2. Move ner_dataset.csv into `./data`
-3. create `./preprocessed_data` directory
 3. `$ cd src && python preprocess.py`
 
 To train the model:

example/named_entity_recognition/src/metrics.py

@@ -27,7 +27,7 @@ def load_obj(name):
     with open(name + '.pkl', 'rb') as f:
         return pickle.load(f)
 
-tag_dict = load_obj("../preprocessed_data/tag_to_index")
+tag_dict = load_obj("../data/tag_to_index")
 not_entity_index = tag_dict["O"]
 
 def classifer_metrics(label, pred):

example/named_entity_recognition/src/ner.py

@@ -34,7 +34,7 @@
 
 parser = argparse.ArgumentParser(description="Deep neural network for multivariate time series forecasting",
                                  formatter_class=argparse.ArgumentDefaultsHelpFormatter)
-parser.add_argument('--data-dir', type=str, default='../preprocessed_data',
+parser.add_argument('--data-dir', type=str, default='../data',
                     help='relative path to input data')
 parser.add_argument('--output-dir', type=str, default='../results',
                     help='directory to save model files to')

example/nce-loss/README.md

@@ -29,7 +29,7 @@ The dataset used in the following examples is [text8](http://mattmahoney.net/dc/
 * word2vec.py: a CBOW word2vec example using nce loss. You need to [download the text8 dataset](#dataset-download) before running this script. Command to start training on CPU (pass -g for training on GPU):
 
 ```
-python word2vec.py
+python wordvec.py
 
 ```