Synthetic dataset generation for the first 5 samples; visualization fixed

change_visualizer.py

@@ -20,8 +20,7 @@ def update_manual_view(diff_idx):
 def update_synthetic_view(diff_idx):
     diff_idx -= 1
     return (df_synthetic.iloc[diff_idx]['annotated_diff'], df_synthetic.iloc[diff_idx]['initial_msg_pred'],
-            df_synthetic.iloc[diff_idx][
-                'get_annotated_diff'],
+            df_synthetic.iloc[diff_idx]['reference'],
             f"https://github.com/{df_synthetic.iloc[diff_idx]['repo']}/commit/{df_synthetic.iloc[diff_idx]['hash']}")
 
 

generate_synthetic_dataset.py

@@ -57,9 +57,16 @@ def generate_synthetic_dataset():
     df['initial_msg_prompt'] = df.apply(generate_prompt_for_row, axis=1)
     initial_messages_pred = []
 
-    for prompt in tqdm(df['initial_msg_prompt']):
-        output = generate_initial_msg(prompt)
-        initial_messages_pred.append(output)
+    for i, prompt in enumerate(tqdm(df['initial_msg_prompt'])):
+        output = None
+
+        if i < 5:
+            while output is None:
+                try:
+                    output = generate_initial_msg(prompt)
+                except:
+                    pass
+        initial_messages_pred.append(output if output is not None else "TBA")
 
     df['initial_msg_pred'] = initial_messages_pred
 

hf_data_loader.py

@@ -19,7 +19,7 @@ def load_full_commit_dataset_as_pandas():
 
 
 def load_synthetic_dataset_as_pandas():
-    load_dataset(config.HF_SYNTHETIC_DATASET_NAME,
-                 split=config.HF_SYNTHETIC_DATASET_SPLIT,
-                 token=config.HF_TOKEN,
-                 cache_dir=config.CACHE_DIR).to_pandas()
+    return load_dataset(config.HF_SYNTHETIC_DATASET_NAME,
+                        split=config.HF_SYNTHETIC_DATASET_SPLIT,
+                        token=config.HF_TOKEN,
+                        cache_dir=config.CACHE_DIR).to_pandas()