Squash Bugs

app.py

@@ -141,5 +141,7 @@ st.subheader('Logged runs')
 #     use_container_width=True
 # )
 
+
+df_runs['spec_version'] = df_runs['spec_version'].astype('int64')
 with st.expander('Show raw run data'):
     st.dataframe(df_runs)

utils.py

@@ -22,10 +22,7 @@ KEYS = ['_step','_timestamp','task','query','reference','challenge','topic','sub
 ABBREV_CHARS = 8
 ENTITY_CHOICES = ('identity', 'hotkey', 'coldkey')
 LOCAL_WANDB_PATH = './data/wandb'
-USERNAME = 'opentensor'
-
-# Set the WANDB_API_KEY environment variable to 'anonymous'
-os.environ["WANDB_API_KEY"] = "anonymous"
+USERNAME = 'login19861986'
 
 # Initialize wandb with anonymous login
 wandb.login(anonymous='must')
@@ -142,8 +139,8 @@ def load_downloaded_runs(time, cols=KEYS):
         'question-answering': 'qa',
     }
 
-    # Runs which do not have a turn field are imputed to be turn zero (single turn)
-    df_all.turn.fillna(0, inplace=True)
+    # # Runs which do not have a turn field are imputed to be turn zero (single turn)
+    # df_all.turn.fillna(0, inplace=True)
 
     df_all.sort_values(by=['_timestamp'], inplace=True)
     # Check if df_all has the task columns
@@ -230,12 +227,15 @@ def download_runs(time, df_vali):
         progress.progress(i/len(df_vali), text=f'{prog_msg}... **downloading** `{os.path.join(*row.run_id)}`')
 
         save_path = f'data/wandb/{row.run_id}.parquet'
+        # Create the directory if it does not exist
+        os.makedirs(os.path.dirname(save_path), exist_ok=True)
+
         if os.path.exists(save_path):
             pbar.set_description(f'>> Skipping {row.run_id!r} because file {save_path!r} already exists')
             continue
 
         try:
-            pbar.set_description(f'* Downloading run {row.run_id!r}', flush=True)
+            pbar.set_description(f'* Downloading run {row.run_id!r}')
             run = api.run(row.run_path)
 
             # By default we just download a subset of events (500 most recent)