Try out new version

data/model.csv

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1b511f072d4b52c995e69aebc2b3f10fff7aa5a860d4c75d8fc7d47ef9194a6d
-size 10292501
+oid sha256:88108b9beaa9863b308ef27805b819cfcc37a14c4c491488a3de9930315a0468
+size 3561158

data/model.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fdc02a0cbeace0ced62cac81640773c36e1325ea59fd85afd12037100d1d1187
-size 7986862
+oid sha256:fbcaffee79f965e3f8eb2820665bf5a926e317a1b1f8401f50a43ff979914f9c
+size 5883182

recommendation-api/learner.py

@@ -1,16 +1,31 @@
-from fastai.collab import load_learner
-from fastai.tabular.all import *
-
-def custom_accuracy(prediction, target):    
-    # set all predictions above 0.95 as true positive (correct prediction)
-    prediction = torch.where(prediction > 0.95, torch.tensor(1.0), prediction)
-    # shape [64, 1] to [64]
-    target = target.squeeze(1)
-    correct = (prediction == target).float()
-    accuracy = correct.sum() / len(target)
-    return accuracy
-
-async def setup_learner(model_filename: str):
-    learn = load_learner(model_filename)
-    learn.dls.device = 'cpu'
+from fastai.collab import load_learner
+from fastai.tabular.all import *
+
+class DotProductBias(Module):
+    def __init__(self, n_users, n_movies, n_factors, y_range=(0,1.1)):
+        self.user_factors = Embedding(n_users, n_factors)
+        self.user_bias = Embedding(n_users, 1)
+        self.movie_factors = Embedding(n_movies, n_factors)
+        self.movie_bias = Embedding(n_movies, 1)
+        self.y_range = y_range
+        
+    def forward(self, x):
+        users = self.user_factors(x[:,0])
+        movies = self.movie_factors(x[:,1])
+        res = (users * movies).sum(dim=1, keepdim=True)
+        res += self.user_bias(x[:,0]) + self.movie_bias(x[:,1])
+        return sigmoid_range(res, *self.y_range)
+
+def custom_accuracy(prediction, target):    
+    # set all predictions above 0.95 as true positive (correct prediction)
+    prediction = torch.where(prediction > 0.95, torch.tensor(1.0), prediction)
+    # shape [64, 1] to [64]
+    target = target.squeeze(1)
+    correct = (prediction == target).float()
+    accuracy = correct.sum() / len(target)
+    return accuracy
+
+async def setup_learner(model_filename: str):
+    learn = load_learner(model_filename)
+    learn.dls.device = 'cpu'
     return learn

recommendation-api/server.py

@@ -7,7 +7,7 @@ from tracks import get_top_tracks_for_user, get_users_with_track_interactions
 from recommender import get_recommendations_for_user
 
 # custom_accuracy needs to be imported to the global namespace for Learner to load
-from learner import setup_learner, custom_accuracy
+from learner import setup_learner, custom_accuracy, DotProductBias
 
 app = FastAPI()
 model_filename = 'data/model.pkl'

recommendation-api/tracks.py

@@ -50,6 +50,6 @@ def predictions_to_tracks(entries_and_predictions):
         track_info = tracks_df[tracks_df['entry'] == entry]
         if not track_info.empty:
             track_dict = track_info.to_dict('records')[0]
-            track_dict['score'] = score.astype(str)
+            track_dict['score'] = score[0].astype(str)
             tracks.append(track_dict)
     return tracks