Chenhui/update diagram (#172)

* update multiround training diagram

* minor change

* update diagram and minor change

Former-commit-id: db07b00674

examples/grocery_sales/python/00_quick_start/autoarima_single_round.ipynb

@@ -1131,7 +1131,7 @@
     "```\n",
     "on your local machine before accessing the dashboard locally. Below is a snapshot of the Ray dashboard during a previous run of the notebook.\n",
     "\n",
-    "<img src=\"https://user-images.githubusercontent.com/20047467/77698387-e8304600-6f86-11ea-9b47-f456cc898c78.png\">"
+    "<img src=\"https://user-images.githubusercontent.com/20047467/77698387-e8304600-6f86-11ea-9b47-f456cc898c78.png\">\n"
    ]
   },
   {

examples/grocery_sales/python/02_model/lightgbm_multi_round.ipynb

@@ -302,9 +302,9 @@
    "source": [
     "## Model Training\n",
     "\n",
-    "We then perform a multi-round training by fitting a LightGBM model using the training data in each forecast round. After the models are trained, we apply them to generate forecasts for the target weeks. The paradigm of model training and evaluation is illustrated in the following figure\n",
+    "We then perform a multi-round training by fitting a LightGBM model using the training data in each forecast round. After the models are trained, we apply them to generate forecasts for the target weeks. The paradigm of model training and testing is illustrated in the following diagram\n",
     "\n",
-    "<img src=\"https://user-images.githubusercontent.com/20047467/74973251-1acba980-53f1-11ea-8f7f-e03238a0daf2.png\" width=\"600\" height=\"320\">"
+    "<img src=\"https://user-images.githubusercontent.com/20047467/77784194-84faee00-7030-11ea-83ee-6e2c33eb2434.png\">"
    ]
   },
   {