diff --git a/chapter01/chapter01.py b/chapter01/chapter01.py
index 3c4488b..e37a3fa 100644
--- a/chapter01/chapter01.py
+++ b/chapter01/chapter01.py
@@ -79,8 +79,6 @@ def is_cat(x): return x[0].isupper()
 
 '''
 
-from fastai.tabular.all import *
-
 path = untar_data(URLs.ADULT_SAMPLE)
 
 dls = TabularDataLoaders.from_csv(path/'adult.csv', path=path, y_names="salary",
@@ -95,8 +93,6 @@ def is_cat(x): return x[0].isupper()
 
 # Exercise 4: Predict number vs categories
 
-from fastai.collab import *
-
 path = untar_data(URLs.ML_SAMPLE)
 dls = CollabDataLoaders.from_csv(path/'ratings.csv')
 learn = collab_learner(dls, y_range=(0.5,5.5)) # Indicate target range
diff --git a/chapter01/white-cat.jpeg b/chapter01/white-cat.jpeg
new file mode 100644
index 0000000..eb1a6b4
Binary files /dev/null and b/chapter01/white-cat.jpeg differ
diff --git a/chapter02/chapter02.py b/chapter02/chapter02.py
index 2e5f405..8285cf2 100644
--- a/chapter02/chapter02.py
+++ b/chapter02/chapter02.py
@@ -1,62 +1,74 @@
+! [ -e /content ] && pip install -Uqq fastbook
+import fastbook
+fastbook.setup_book()
 
-key = 'XXX'
-key = os.environ['AZURE_SEARCH_KEY']
+from fastbook import *
+from fastai.vision.widgets import *
 
+key = os.environ.get('AZURE_SEARCH_KEY', 'X')
 
 search_images_bing
 
 results = search_images_bing(key, 'grizzly bear')
-ims = results.attrgot('content_url')
+ims = results.attrgot('contentUrl')
 len(ims)
 
+ims = ['http://3.bp.blogspot.com/-S1scRCkI3vY/UHzV2kucsPI/AAAAAAAAA-k/YQ5UzHEm9Ss/s1600/Grizzly%2BBear%2BWildlife.jpg']
 dest = 'images/grizzly.jpg'
 download_url(ims[0], dest)
 im = Image.open(dest)
 im.to_thumb(128,128)
 
+# Download all the URLs for each of the search terms, each on a separate folder
+
 bear_types = 'grizzly','black','teddy'
 path = Path('bears')
-if not path.exists(): path.mkdir()
-for o in bear_types: dest = (path/o)
-            dest.mkdir(exist_ok=True)
-            results = search_images_bing(key, f'{o} bear')
-            download_images(dest, urls=results.attrgot('content_url'))
-            
+
+if not path.exists():
+    path.mkdir()
+    for o in bear_types:
+        dest = (path/o)
+        dest.mkdir(exist_ok=True)
+        results = search_images_bing(key, f'{o} bear')
+        download_images(dest, urls=results.attrgot('contentUrl'))
+
 fns = get_image_files(path)
 fns
 
-failed = verify_images(fns)
-
-failed
+# Check and delete corrupt images
 
+failed = verify_images(fns)
 failed.map(Path.unlink);
 
-
 class DataLoaders(GetAttr):
     def __init__(self, *loaders): self.loaders = loaders
-    def __getitem__(self, i): return self.loaders[i] train,valid = add_props(lambda i,self: self[i])
+    def __getitem__(self, i): return self.loaders[i]
+    train,valid = add_props(lambda i,self: self[i])
 
 bears = DataBlock(
-        blocks=(ImageBlock, CategoryBlock),
-        get_items=get_image_files,
-        splitter=RandomSplitter(valid_pct=0.2, seed=42),
-        get_y=parent_label,
-        item_tfms=Resize(128))
-
-blocks=(ImageBlock, CategoryBlock)
 
-get_items=get_image_files
+    # Types for independent and dependent variables
+    # X. Independent = set of images, prediction source
+    # Y. Dependent = categories, target prediction
+    blocks=(ImageBlock, CategoryBlock),
 
+    # Underlying items, file paths
+    get_items=get_image_files,
 
-splitter=RandomSplitter(valid_pct=0.2, seed=42)
+    # Random splitter with fixed seed
+    splitter=RandomSplitter(valid_pct=0.2, seed=42),
 
-get_y=parent_label
+    # Set labels using the file name
+    get_y=parent_label,
 
-item_tfms=Resize(128)
+    # Item transforms. Returns a DataBlock object
+    item_tfms=Resize(128))
 
+# Data source
+# Includes validation and training
+# Mini-batch of 64 items at a time in a single tensor
 dls = bears.dataloaders(path)
 
-
 dls.valid.show_batch(max_n=4, nrows=1)
 
 bears = bears.new(item_tfms=Resize(128, ResizeMethod.Squish))
@@ -67,20 +79,98 @@ def __getitem__(self, i): return self.loaders[i] train,valid = add_props(lambda
 dls = bears.dataloaders(path)
 dls.valid.show_batch(max_n=4, nrows=1)
 
+# Data augmentation. Randomized cropped image
 bears = bears.new(item_tfms=RandomResizedCrop(128, min_scale=0.3))
 dls = bears.dataloaders(path)
+
+# Same image, different versions of the crop transform
 dls.train.show_batch(max_n=4, nrows=1, unique=True)
 
+# Data augmentation
 bears = bears.new(item_tfms=Resize(128), batch_tfms=aug_transforms(mult=2))
 dls = bears.dataloaders(path)
 dls.train.show_batch(max_n=8, nrows=2, unique=True)
 
+# Standard size 244x244 pixel
+# Train bear classifier
 bears = bears.new(
-        item_tfms=RandomResizedCrop(224, min_scale=0.5),
-        batch_tfms=aug_transforms())
+    item_tfms=RandomResizedCrop(224, min_scale=0.5),
+    batch_tfms=aug_transforms())
 dls = bears.dataloaders(path)
 
-earn = cnn_learner(dls, resnet18, metrics=error_rate)
+# Create learner and fine-tune
+learn = vision_learner(dls, resnet18, metrics=error_rate)
 learn.fine_tune(4)
 
+# Confusion matrix to measure accuracy
+# Diagonal correct classification
+
+interp = ClassificationInterpretation.from_learner(learn)
+interp.plot_confusion_matrix()
+
+interp.plot_top_losses(5, nrows=1)
+
+#hide_output
+cleaner = ImageClassifierCleaner(learn)
+cleaner
+
+# Unlink all images selected for deletion
+for idx in cleaner.delete(): cleaner.fns[idx].unlink()
+
+# Different category
+for idx,cat in cleaner.change(): shutil.move(str(cleaner.fns[idx]), path/cat)
+
+learn.export()
+
+# Check if file exists
+path = Path()
+path.ls(file_exts='.pkl')
+
+# Simulating inference
+learn_inf = load_learner(path/'export.pkl')
+
+# Filename to predict
+learn_inf.predict('images/grizzly.jpg')
+
+# Vocab, or stored list of all possible categories
+learn_inf.dls.vocab
+
+# Ipywidget
+btn_upload = widgets.FileUpload()
+btn_upload
+
+img = PILImage.create(btn_upload.data[-1])
+
+# Display
+out_pl = widgets.Output()
+out_pl.clear_output()
+with out_pl: display(img.to_thumb(128,128))
+out_pl
+
+# Get predictions
+pred,pred_idx,probs = learn_inf.predict(img)
+
+# Display
+lbl_pred = widgets.Label()
+lbl_pred.value = f'Prediction: {pred}; Probability: {probs[pred_idx]:.04f}'
+lbl_pred
+
+# Classification button
+btn_run = widgets.Button(description='Classify')
+btn_run
+
+def on_click_classify(change):
+    img = PILImage.create(btn_upload.data[-1])
+    out_pl.clear_output()
+    with out_pl: display(img.to_thumb(128,128))
+    pred,pred_idx,probs = learn_inf.predict(img)
+    lbl_pred.value = f'Prediction: {pred}; Probability: {probs[pred_idx]:.04f}'
+
+btn_run.on_click(on_click_classify)
+
+# Reset values
+btn_upload = widgets.FileUpload()
+
+VBox([widgets.Label('Select your bear!'),
+      btn_upload, btn_run, out_pl, lbl_pred])
 
diff --git a/chapter02/teddy-bear.jpg b/chapter02/teddy-bear.jpg
new file mode 100644
index 0000000..77f2aad
Binary files /dev/null and b/chapter02/teddy-bear.jpg differ