Update B_Model_Tutorial_4_Hidden_Markov_Models.ipynb (#1022)

docs/tutorials/B_Model_Tutorial_4_Hidden_Markov_Models.ipynb

@@ -210,7 +210,7 @@
    "id": "66fb5e21",
    "metadata": {},
    "source": [
-    "It looks like it successfully identified a CG island in the middle (the long stretch of 0's) and another shorter one at the end. More importantly, the model wasn't tricked into thinking that every CG or even pair of CGs was an island. It required many C's and G's to be part of a longer stretch to identify that region as an island. Naturally, the balance of the transition and emission probabilities will heavily influence what regions are detected.\n",
+    "It looks like it successfully identified a CG island in the middle (the long stretch of 1's) and another shorter one at the end. More importantly, the model wasn't tricked into thinking that every CG or even pair of CGs was an island. It required many C's and G's to be part of a longer stretch to identify that region as an island. Naturally, the balance of the transition and emission probabilities will heavily influence what regions are detected.\n",
     "\n",
     "Let's say, though, that we want to get rid of that CG island prediction at the end because we don't believe that real islands can occur at the end of the sequence. We can take care of this by adding in an explicit end state that only the non-island hidden state can get to. We enforce that the model has to end in the end state, and if only the non-island state gets there, the sequence of hidden states must end in the non-island state. Here's how:"
    ]