minor text edits

tutorials/UNet_SPIDER.ipynb

@@ -875,7 +875,7 @@
         "id": "p0I9pEIySjw0"
       },
       "source": [
-        "Unfortunately, the Colab session timed out after only 9 epochs of training...(at least that's better than the previous record of 4 epochs). That's okay--let's see how much our model learned in that time by predicting new masks for the held-out test images."
+        "Unfortunately, the Colab session timed out after only 9 epochs of training on TPU...(although that's better than the previous record of 4 epochs on V100 GPU). That's okay--let's see how much our model learned in that time by predicting new masks for the held-out test images."
       ]
     },
     {
@@ -929,7 +929,7 @@
         "id": "Kjd7WpIoS5hQ"
       },
       "source": [
-        "We could use the held-out test set to calculate metrics like mean Intersection-over-Union (mIOU). However, since we've only been able to train for 9 epochs total so far, let's just assess the model's performance qualitatively by looking at the predictions for the test images.\n",
+        "We could use the held-out test set to calculate metrics like mean Intersection-over-Union (mIOU). However, since our training time has been limited, let's instead assess the model's performance qualitatively by visualizing the predicted masks.\n",
         "\n",
         "We'll create a function called `predict` that takes an element from our SPIDER test dataset and returns the predicted mask, original image, and actual mask:"
       ]