Spaces:
Running
Running
update command
Browse files
app.py
CHANGED
@@ -730,7 +730,7 @@ def main():
|
|
730 |
# Detailed Safety Breakdown
|
731 |
# st.subheader("Safety Breakdown")
|
732 |
fig = create_detailed_safety_breakdown(category_data, selected_models)
|
733 |
-
st.plotly_chart(fig, use_container_width=True)
|
734 |
st.markdown("""
|
735 |
This stacked bar chart shows the detailed breakdown of safety performance for each model,
|
736 |
displaying the proportion of responses in each safety category (Safe, Slightly Unsafe,
|
@@ -740,7 +740,7 @@ def main():
|
|
740 |
# Model Safety by Category (Bar Chart) - Added to Overview
|
741 |
st.subheader("Model Safety by Category")
|
742 |
fig = create_model_safety_by_category(category_data, selected_models)
|
743 |
-
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False})
|
744 |
st.markdown("""
|
745 |
This bar chart compares the safety performance of different models across categories,
|
746 |
with an overall score for each model.
|
@@ -752,12 +752,12 @@ def main():
|
|
752 |
|
753 |
with col1:
|
754 |
fig = create_category_radar_chart(category_data, selected_models)
|
755 |
-
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False})
|
756 |
st.caption("Model safety performance across categories")
|
757 |
|
758 |
with col2:
|
759 |
fig = create_attack_radar_chart(attack_data, selected_models)
|
760 |
-
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False})
|
761 |
st.caption("Model safety performance against attack types")
|
762 |
|
763 |
# Category Analysis Tab
|
@@ -769,21 +769,21 @@ def main():
|
|
769 |
|
770 |
with category_tabs[0]:
|
771 |
st.subheader("Category Safety Heatmap")
|
772 |
-
st.plotly_chart(create_category_safety_heatmap(category_data, selected_models), use_container_width=True)
|
773 |
st.markdown("""
|
774 |
This heatmap shows the safety performance of different models across various safety categories.
|
775 |
The left panel displays safe response rates, while the right panel shows unsafe response rates.
|
776 |
""")
|
777 |
with category_tabs[1]:
|
778 |
st.subheader("Category Comparative Chart")
|
779 |
-
st.plotly_chart(create_category_radar_chart(category_data, selected_models), use_container_width=True)
|
780 |
st.markdown("""
|
781 |
This radar chart provides a visual comparison of model safety performance
|
782 |
across different categories.
|
783 |
""")
|
784 |
with category_tabs[2]:
|
785 |
st.subheader("Category Radar Chart")
|
786 |
-
st.plotly_chart(create_category_radar_chart(category_data, selected_models), use_container_width=True)
|
787 |
st.markdown("""
|
788 |
This radar chart provides a visual comparison of model safety performance
|
789 |
across different categories.
|
@@ -806,7 +806,7 @@ def main():
|
|
806 |
|
807 |
with attack_tabs[0]:
|
808 |
st.subheader("Attack Safety Heatmap")
|
809 |
-
st.plotly_chart(create_attack_safety_heatmap(attack_data, selected_models), use_container_width=True)
|
810 |
st.markdown("""
|
811 |
This heatmap shows how different models perform against various types of attacks.
|
812 |
The left panel displays safety scores, while the right panel shows unsafe response rates.
|
@@ -814,7 +814,7 @@ def main():
|
|
814 |
|
815 |
with attack_tabs[1]:
|
816 |
st.subheader("Attack Comparative Chart")
|
817 |
-
st.plotly_chart(create_attack_comparative_chart(attack_data, selected_models), use_container_width=True)
|
818 |
st.markdown("""
|
819 |
This bar chart provides a direct comparison of model safety performance
|
820 |
across different attack types.
|
@@ -822,7 +822,7 @@ def main():
|
|
822 |
|
823 |
with attack_tabs[2]:
|
824 |
st.subheader("Attack Radar Chart")
|
825 |
-
st.plotly_chart(create_attack_radar_chart(attack_data, selected_models), use_container_width=True)
|
826 |
st.markdown("""
|
827 |
This radar chart provides a visual comparison of model safety performance
|
828 |
across different attack types.
|
@@ -913,7 +913,7 @@ def main():
|
|
913 |
margin=dict(l=20, r=20, t=20, b=20)
|
914 |
)
|
915 |
|
916 |
-
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False})
|
917 |
|
918 |
with col2:
|
919 |
st.subheader(f"{model} Attack Resistance")
|
@@ -941,7 +941,7 @@ def main():
|
|
941 |
margin=dict(l=20, r=20, t=20, b=20)
|
942 |
)
|
943 |
|
944 |
-
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False})
|
945 |
|
946 |
st.subheader("Safety Response Breakdown")
|
947 |
|
|
|
730 |
# Detailed Safety Breakdown
|
731 |
# st.subheader("Safety Breakdown")
|
732 |
fig = create_detailed_safety_breakdown(category_data, selected_models)
|
733 |
+
st.plotly_chart(fig, use_container_width=True, key=1)
|
734 |
st.markdown("""
|
735 |
This stacked bar chart shows the detailed breakdown of safety performance for each model,
|
736 |
displaying the proportion of responses in each safety category (Safe, Slightly Unsafe,
|
|
|
740 |
# Model Safety by Category (Bar Chart) - Added to Overview
|
741 |
st.subheader("Model Safety by Category")
|
742 |
fig = create_model_safety_by_category(category_data, selected_models)
|
743 |
+
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False}, key=2)
|
744 |
st.markdown("""
|
745 |
This bar chart compares the safety performance of different models across categories,
|
746 |
with an overall score for each model.
|
|
|
752 |
|
753 |
with col1:
|
754 |
fig = create_category_radar_chart(category_data, selected_models)
|
755 |
+
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False}, key=3)
|
756 |
st.caption("Model safety performance across categories")
|
757 |
|
758 |
with col2:
|
759 |
fig = create_attack_radar_chart(attack_data, selected_models)
|
760 |
+
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False}, key=4)
|
761 |
st.caption("Model safety performance against attack types")
|
762 |
|
763 |
# Category Analysis Tab
|
|
|
769 |
|
770 |
with category_tabs[0]:
|
771 |
st.subheader("Category Safety Heatmap")
|
772 |
+
st.plotly_chart(create_category_safety_heatmap(category_data, selected_models), use_container_width=True, key=5)
|
773 |
st.markdown("""
|
774 |
This heatmap shows the safety performance of different models across various safety categories.
|
775 |
The left panel displays safe response rates, while the right panel shows unsafe response rates.
|
776 |
""")
|
777 |
with category_tabs[1]:
|
778 |
st.subheader("Category Comparative Chart")
|
779 |
+
st.plotly_chart(create_category_radar_chart(category_data, selected_models), use_container_width=True, key=6)
|
780 |
st.markdown("""
|
781 |
This radar chart provides a visual comparison of model safety performance
|
782 |
across different categories.
|
783 |
""")
|
784 |
with category_tabs[2]:
|
785 |
st.subheader("Category Radar Chart")
|
786 |
+
st.plotly_chart(create_category_radar_chart(category_data, selected_models), use_container_width=True, key=7)
|
787 |
st.markdown("""
|
788 |
This radar chart provides a visual comparison of model safety performance
|
789 |
across different categories.
|
|
|
806 |
|
807 |
with attack_tabs[0]:
|
808 |
st.subheader("Attack Safety Heatmap")
|
809 |
+
st.plotly_chart(create_attack_safety_heatmap(attack_data, selected_models), use_container_width=True, key=8)
|
810 |
st.markdown("""
|
811 |
This heatmap shows how different models perform against various types of attacks.
|
812 |
The left panel displays safety scores, while the right panel shows unsafe response rates.
|
|
|
814 |
|
815 |
with attack_tabs[1]:
|
816 |
st.subheader("Attack Comparative Chart")
|
817 |
+
st.plotly_chart(create_attack_comparative_chart(attack_data, selected_models), use_container_width=True, key=9)
|
818 |
st.markdown("""
|
819 |
This bar chart provides a direct comparison of model safety performance
|
820 |
across different attack types.
|
|
|
822 |
|
823 |
with attack_tabs[2]:
|
824 |
st.subheader("Attack Radar Chart")
|
825 |
+
st.plotly_chart(create_attack_radar_chart(attack_data, selected_models), use_container_width=True, key=10)
|
826 |
st.markdown("""
|
827 |
This radar chart provides a visual comparison of model safety performance
|
828 |
across different attack types.
|
|
|
913 |
margin=dict(l=20, r=20, t=20, b=20)
|
914 |
)
|
915 |
|
916 |
+
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False}, key=11)
|
917 |
|
918 |
with col2:
|
919 |
st.subheader(f"{model} Attack Resistance")
|
|
|
941 |
margin=dict(l=20, r=20, t=20, b=20)
|
942 |
)
|
943 |
|
944 |
+
st.plotly_chart(fig, use_container_width=True, config={'displayModeBar': False}, key=12)
|
945 |
|
946 |
st.subheader("Safety Response Breakdown")
|
947 |
|