Added examples instructions and fixed notebook typos

Author: marshka

examples/README.md

@@ -0,0 +1,40 @@
+# Running the examples
+
+In this folder, we provide example scripts on how to use TSL for running 
+consistent and reproducible experiments on neural spatiotemporal forecasting 
+and imputation.
+
+The experiment code and its configuration are handled by the [`Experiment`](https://torch-spatiotemporal.readthedocs.io/en/latest/modules/experiment.html) 
+object in `tsl.experiment`, which makes use of [Hydra](https://hydra.cc/) as a 
+configuration manager. Please refer to [Hydra's documentation](https://hydra.cc/docs/intro/)
+for more usage information.
+
+In both subfolders, you will find an example script and a `config` directory.
+We now show an example of how to launch a traffic forecasting experiment (the 
+same procedure applies to the imputation experiment).
+
+Running a forecasting experiment using one of the baseline models on a provided 
+benchmark dataset is as simple as that:
+
+```bash
+python forecasting/run_traffic_experiment.py config=dcrnn dataset.name=la
+```
+
+Under the hood, we are telling the program to run the traffic experiment using 
+the `config/traffic/dcrnn.yaml` config, which is a particular instance of the 
+`config/traffic/default.yaml` config. In practice, any parameter not defined in 
+the primary `dcrnn.yaml` config file is inherited from the fallback config 
+`default.yaml`.
+
+Note that in this latter config, the rows
+```yaml
+dataset:
+  name: ???
+```
+indicate that the name of the dataset is required in order to run the 
+experiment. By passing as argument `dataset.name=la` we run the forecasting 
+experiment on the [MetrLA](https://torch-spatiotemporal.readthedocs.io/en/latest/modules/datasets_in_tsl.html#tsl.datasets.MetrLA)
+dataset.
+
+Check the code in the script file to see other possible options for the 
+configuration arguments!