Update app.py

app.py

@@ -8,7 +8,6 @@ import re
 
 def load_leaderboard():
     # Load validation / test CSV files
-
     results_csv_files = {
         'imagenet': 'https://raw.githubusercontent.com/huggingface/pytorch-image-models/main/results/results-imagenet.csv',
         'real': 'https://raw.githubusercontent.com/huggingface/pytorch-image-models/main/results/results-imagenet-real.csv',
@@ -32,6 +31,7 @@ def load_leaderboard():
     dataframes = {name: pd.read_csv(url) for name, url in results_csv_files.items()}
     bench_dataframes = {name: pd.read_csv(url) for name, url in benchmark_csv_files.items()}
     bench_dataframes = {name: df for name, df in bench_dataframes.items() if 'infer_gmacs' in df.columns}
+    print(bench_dataframes.keys())
 
     # Clean up dataframes
     remove_column_names = ["top1_err", "top5_err", "top1_diff", "top5_diff", "rank_diff", "param_count"]
@@ -88,6 +88,7 @@ def load_leaderboard():
 
     return merged_dataframes
 
+
 REGEX_PREFIX = "re:"
 
 def auto_match(pattern, text):
@@ -99,36 +100,34 @@ def auto_match(pattern, text):
         except re.error:
             # If it's an invalid regex, return False
             return False
-    
+
     # Check if it's a wildcard pattern
     elif any(char in pattern for char in ['*', '?']):
         return fnmatch.fnmatch(text.lower(), pattern.lower())
-    
+
     # If not regex or wildcard, use fuzzy matching
     else:
         return fuzz.partial_ratio(
             pattern.lower(), text.lower(), score_cutoff=90) > 0
 
-
 def filter_leaderboard(df, model_name, sort_by):
     if not model_name:
         return df.sort_values(by=sort_by, ascending=False)
-    
+
     mask = df['model'].apply(lambda x: auto_match(model_name, x))
     filtered_df = df[mask].sort_values(by=sort_by, ascending=False)
-    
-    return filtered_df
 
+    return filtered_df
 
-def create_scatter_plot(df, x_axis, y_axis, model_filter, highlight_filter):
+def create_scatter_plot(df, x_axis, y_axis, model_filter, highlight_filter, log_x, log_y):
     selected_color = 'orange'
-    
+
     fig = px.scatter(
         df,
         x=x_axis,
         y=y_axis,
-        log_x=True,
-        log_y=True,
+        log_x=log_x,
+        log_y=log_y,
         hover_data=['model'],
         trendline='ols',
         trendline_options=dict(log_x=True, log_y=True),
@@ -144,19 +143,18 @@ def create_scatter_plot(df, x_axis, y_axis, model_filter, highlight_filter):
         legend_labels[False] = f'{model_filter or "all models"}'
     else:
         legend_labels[False] = f'{model_filter or "all models"}'
-    
+
     # Update legend
     for trace in fig.data:
         if isinstance(trace.marker.color, str):  # This is for the scatter traces
             trace.name = legend_labels.get(trace.marker.color == selected_color, '')
-    
+
     fig.update_layout(
         showlegend=True,
         legend_title_text='Model Selection'
     )
-    
-    return fig
 
+    return fig
 
 # Load the leaderboard data
 merged_dataframes = load_leaderboard()
@@ -217,7 +215,6 @@ def update_leaderboard_and_plot(
 
     return display_df, fig
 
-
 with gr.Blocks(title="The timm Leaderboard") as app:
     gr.HTML("<center><h1>The timm (PyTorch Image Models) Leaderboard</h1></center>")
     gr.HTML("<p>This leaderboard is based on the results of the models from <a href='https://github.com/huggingface/pytorch-image-models'>timm</a>.</p>")
@@ -265,4 +262,4 @@ with gr.Blocks(title="The timm Leaderboard") as app:
     log_y.change(update_leaderboard_and_plot, inputs=inputs, outputs=outputs)
     update_btn.click(update_leaderboard_and_plot, inputs=inputs, outputs=outputs)
 
-app.launch()
+app.launch(debug=True)