conclusion is done

deep-learning-intro-for-hep/xx-conclusion.md

@@ -1 +1,11 @@
 # Conclusion
+
+I hope you enjoyed this introduction to deep learning for particle physicists! I especially hope that it has dispelled the aura of mystery surrounding ML, so that you can approach ML problems with the same confidence as a simple fit. Maybe (likely) your deep neural networks won't work the first time you try to train them, but now you should be able to pull them apart and understand each piece and how they fit together to diagnose your problems, just as you would with any other broken software. The _one thing_ that's fundamentally different between your ML projects and all your other code is that the ML depends on a huge number of numeric parameters that are optimized by function minimization. The rest is mostly array manipulation.
+
+If you see any problems in this book, you can report them as [GitHub Issues](https://github.com/hsf-training/deep-learning-intro-for-hep/issues).
+
+If you'd like to contribute or fix something, you can open a [pull request](https://github.com/hsf-training/deep-learning-intro-for-hep/pulls).
+
+The sequence of sections leading up to the [main project](20-main-project.md) probably won't change, but you or I might add sections after it, since these are introducing readers to the variety of deep learning techniques that are useful in HEP.
+
+Cheers!