Merge branch 'master' of https://github.com/jmschrei/pomegranate

tutorials/Tutorial_7_Parallelization.ipynb

@@ -16,7 +16,7 @@
     "6. log_probability\n",
     "7. probability\n",
     "\n",
-    "These wrappers are all fairly simple, just chunking the data and passing them into the appropriate function. Any of the models in pomegranate can use them to varying degrees of efficiency. These functions work best when you have large amounts of data requiring several seconds (preferably more than 10) to process on a single processor. This is because it takes some time to set up the worker pool, and the parallelization is still being imporved.\n",
+    "These wrappers are all fairly simple, just chunking the data and passing them into the appropriate function. Any of the models in pomegranate can use them to varying degrees of efficiency. These functions work best when you have large amounts of data requiring several seconds (preferably more than 10) to process on a single processor. This is because it takes some time to set up the worker pool, and the parallelization is still being improved.\n",
     "\n",
     "In this tutorial, I will show you how you can use the parallelization wrappers to speed up fitting and predictions in models of increasing complexity. First we'll look at a simple multivariate Gaussian mixture model, and compare its performance to sklearn. Then we'll look at the first layer of model stacking that pomegranate supports by showing a hidden Markov model with GMM emissions, known as a GMM-HMM. Lastly, we'll go to the second layer of model stacking by showing a mixture of GMM-HMMs, which is a mixture model stacked inside a hidden markov model stacked inside a mixture model. These can all utilize the build-in parallelization functions that pomegranate has.\n",
     "\n",