QBGPT V1.0

app.py

@@ -32,18 +32,18 @@ qbgpt = QBGPT(input_vocab_size = input_size,
                     diag_masks = False,
                     to_pred_size = moves_to_pred)
 
-qbgpt.load_weights("app/assets/model_mediumv2/QBGPT")
+qbgpt.load_weights("assets/model_mediumv2/QBGPT")
 
 
-qb_tok = tokenizer(moves_index="./app/assets/moves_index.parquet",
-                   play_index="./app/assets/plays_index.parquet",
-                   positions_index="./app/assets/positions_index.parquet",
-                   scrimmage_index="./app/assets/scrimmage_index.parquet",
-                   starts_index="./app/assets/starts_index.parquet",
-                   time_index="./app/assets/time_index.parquet",
+qb_tok = tokenizer(moves_index="./assets/moves_index.parquet",
+                   play_index="./assets/plays_index.parquet",
+                   positions_index="./assets/positions_index.parquet",
+                   scrimmage_index="./assets/scrimmage_index.parquet",
+                   starts_index="./assets/starts_index.parquet",
+                   time_index="./assets/time_index.parquet",
                    window_size=20)
 
-with open('./app/assets/ref.json', 'r') as fp:
+with open('./assets/ref.json', 'r') as fp:
     ref_json = json.load(fp)
     
 def convert_numpy(d):
@@ -51,7 +51,7 @@ def convert_numpy(d):
 
 ref_json = {int(k):convert_numpy(v) for k,v in ref_json.items()}
 
-ref_df = pd.read_json("./app/assets/ref_df.json")
+ref_df = pd.read_json("./assets/ref_df.json")
 
 
 

pages.py

@@ -63,7 +63,7 @@ def qb_gpt_page(ref_df, ref, tokenizer, model):
         plot_true = pd.DataFrame(step1_true)
         
         fig_gen = px.line(plot, x="input_ids_x", y="input_ids_y", animation_frame="pos_ids", color="OffDef", symbol="ids",
-                        text="position_ids", title="Player Trajectories Over Time", line_shape="linear",
+                        text="position_ids", title="Generated players' trajectories Over Time", line_shape="linear",
                         range_x=[0, 140], range_y=[0, 60], # Set X and Y axis ranges
                         render_mode="svg")  # Render mode for smoother lines
 
@@ -73,7 +73,7 @@ def qb_gpt_page(ref_df, ref, tokenizer, model):
         st.plotly_chart(fig_gen)
         
         fig_true = px.line(plot_true, x="input_ids_x", y="input_ids_y", animation_frame="pos_ids", color="OffDef", symbol="ids",
-                    text="position_ids", title="Player Trajectories Over Time",
+                    text="position_ids", title="True players' trajectories Over Time",
                     range_x=[0, 140], range_y=[0, 60], # Set X and Y axis ranges
                     line_shape="linear",  # Draw lines connecting points
                     render_mode="svg")  # Render mode for smoother lines
@@ -106,7 +106,7 @@ def contacts_and_disclaimers():
         qb_gpt_transf = """
         At the heart of QB-GPT lies the cutting-edge Transformer model, a deep learning architecture known for its prowess in understanding sequential data. It doesn't just create plays; it understands the game at a granular level, taking into account player positions, game situations, and historical data. It relies on the same conceptual approach behind the now famous "GPT" model of OpenAI. It's the playbook of the future, driven by the technology of tomorrow.
         
-        A more detailed blogpost about the model QB-GPT can be found [here](link)
+        A more detailed blogpost about the model QB-GPT can be found [here](https://medium.com/@sam.chaineau/transformers-can-generate-nfl-plays-introducing-qb-gpt-2d40f16a03eb)
         """
         st.markdown(qb_gpt_transf)