fix(example): add notebook examples

README.md

@@ -197,7 +197,7 @@ model = AutoModel.from_config(config, predict_sequence_length=7)
 
 **Build your own model**
 
-<details><summary> Full list of model tfts supported using AutoModel </summary>
+<details><summary> Full list of tfts AutoModel supported </summary>
 
 - rnn
 - tcn
@@ -212,7 +212,7 @@ model = AutoModel.from_config(config, predict_sequence_length=7)
 
 </details>
 
-You could build the custom model based on tfts, especially
+You could build the custom model based on tfts, like
 - add custom-defined embeddings for categorical variables
 - add custom-defined head layers for classification or anomaly task
 

docs/source/tricks.rst

@@ -34,14 +34,14 @@ There is no free launch, and it's impossible to forecast the future. So we shoul
 
 	we can train different models from different scale
 
+* module usage
 
-Multi-steps prediction strategy
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+    Be careful to use the layer like `Dropout` or `BatchNorm` for regression task
 
-* multi models
 
-* add a hidden-sizes dense layer at last
+* Multi-steps prediction strategy
 
-* encoder-decoder structure
-
-* encoder-forecasting structure
+    * multi models for single variable prediction
+    * add a hidden-sizes dense layer at last
+    * encoder-decoder structure
+    * encoder-forecasting structure

docs/source/tutorials.rst

@@ -7,7 +7,7 @@ Tutorials
 
     <a class="github-button" href="https://github.com/LongxingTan/Time-series-prediction" data-icon="octicon-star" data-size="large" data-show-count="true" aria-label="Star LongxingTan/Time-series-prediction on GitHub">GitHub</a>
 
-The following tutorials can be also found as `notebooks on GitHub <https://github.com/longxingtan/time-series-prediction/tree/master/notebooks>`_.
+The following tutorials can be also found as `notebooks on GitHub <https://github.com/longxingtan/time-series-prediction/tree/master/examples/notebooks>`_.
 
 .. _prepare_data:
 
@@ -51,6 +51,10 @@ input 1
 Train the models
 -----------------
 
+- Multi-GPU training with `tf.distribute <https://www.tensorflow.org/guide/keras/distributed_training>`_
+- Mixed precision with `tf.keras.mixed_precision <https://www.tensorflow.org/guide/mixed_precision>`_
+
+
 .. code-block:: python
 
     import tensorflow as tf
@@ -90,7 +94,8 @@ Change the model parameters. If you want touch more parameters in model config,
         "rnn_size": 128,
         "dense_size": 128,
     }
-    model = AutoModel('rnn', custom_model_config=custom_model_config)
+    config.update(custom_model_config)
+    model = AutoModel('rnn', config=config)
 
 
 Multi-variables and multi-steps prediction
@@ -106,8 +111,8 @@ Multi-variables and multi-steps prediction
 	print(config)
 
 	config.update({
-	"rnn_size": 128,
-    	"dense_size": 128,
+	    "rnn_size": 128,
+        "dense_size": 128,
 	})
 	print(config)
 
@@ -154,7 +159,7 @@ Set up the custom-defined head layer to do the classification task or anomaly de
 Custom-defined trainer
 ----------------------------------------
 
-You could use tfts trainer, a custom trainer or use keras to train directly.
+You could use `tfts trainer`, `custom trainer` or use keras to train directly.
 
 .. code-block:: python
 

examples/README.md

@@ -1,10 +1,16 @@
 # TFTS examples
 
+**Basic Usage**
 - [time series prediction](./run_prediction_simple.py)
 - [time series anomaly detection](./run_anomaly.py)
 
 
+## Notebooks
+- [single step prediction](https://nbviewer.org/github/LongxingTan/Time-series-prediction/blob/master/examples/notebooks/demo_single_step_prediction.ipynb)
+
+
 ## More examples
 
+**Multiple steps prediction**
 - [TFTS-Bert](https://github.com/LongxingTan/KDDCup2022-Baidu) wins the **3rd place** in KDD Cup 2022 wind power forecasting
 - [TFTS-Seq2seq](https://github.com/LongxingTan/Data-competitions/tree/master/tianchi-enso-prediction) wins the **4th place** in Tianchi ENSO prediction 2021