Get video to work and improve README

README.md

@@ -5,21 +5,22 @@ Interpretable Machine Learning for high-resolution astronomical spectroscopy.
 <a href="https://blase.readthedocs.io/en/latest/"><img src="https://img.shields.io/badge/Read-the%20docs-blue"></a>
 <a href="https://ui.adsabs.harvard.edu/abs/2022ApJ...941..200G/abstract"><img src="https://img.shields.io/badge/Paper-Gully--Santiago & Morley (2022)-green"></a>
 
-## Massively scalable
+## _Massively scalable_
 
 By using autodiff, we can fit over 40,000 spectral lines simultaneously.
-<video width="320" autoplay loop muted playsinline>
-
-  <source src="'https://user-images.githubusercontent.com/860227/236579123-1acf7f23-e202-4c97-95d4-fc149ecdd191.mp4" type="video/mp4" />
-  Your browser does not support the video tag.
+<video width="320" src="https://user-images.githubusercontent.com/860227/236579123-1acf7f23-e202-4c97-95d4-fc149ecdd191.mp4" autoplay loop muted playsinline>
 </video>
 
-## Rooted in physics
+## _Rooted in physics_
 
 We first clone a precomputed synthetic spectrum, such as PHOENIX, and then **transfer learn** with data.
 
-## Blazing fast with GPUs
+## _Blazing fast with GPUs_
 
 We achieve $>60 \times$ speedups with NVIDIA GPUs, so training takes minutes instead of hours.
 
+## Get started
+
+Visit our [step-by-step tutorials](https://blase.readthedocs.io/en/latest/tutorials/index.html) or [installation](https://blase.readthedocs.io/en/latest/install.html) pages to get started. We also have [deep dives](https://blase.readthedocs.io/en/latest/deep_dives/index.html#), or you can [read the paper](https://ui.adsabs.harvard.edu/abs/2022ApJ...941..200G/abstract). Have a question or a research project in mind? Open [an Issue](https://github.com/gully/blase/issues) or [email gully](https://gully.github.io/).
+
 Copyright 2020, 2021, 2022, 2023 The Authors