Update Seq2Seq-gait-analysis/README

[escorciav]

Is it reasonable to break this into different sections?

1. Setup and installation
2. Data setup (optional)
3. Using a pre-trained model
4. Training on your own data
5. Replicating SVM baseline

• You may consider adding a "table of content" to ease navigation. Don't forget to use markdown links.

• It's fine if they overlap content as long as it's easier to follow for the user.

• If one step requires the output of a previous step, consider mentioning that after citing the command. It helps to troubleshoot errors.

  For example

  # Training
  
  `python conv_classifier_eval.py`
  
  This generates a `model.npz` file with the learned parameters.
  

  In that way, the user knows that if they don't see model.npz the next command is most likely failing 'cause of that.

• Mention complementary comments after you show the command. For example,

  Train the Seq2Seq encoder-decoder model
  
  `python bidirectional_autoencoder.py`
  
  Using a GPU reduces the time to complete this step.
  

Resources
This short course outlines the mindset that we should follow.