Update utils.py

moved plotting function here

utils.py

@@ -20,4 +20,28 @@ def predict(model, sequence):
     probabilities = F.softmax(output.logits, dim=-1)
     predicted_label = torch.argmax(probabilities, dim=-1)
     confidence = probabilities.max().item() * 0.85
-    return predicted_label.item(), confidence
+    return predicted_label.item(), confidence
+
+def plot_prediction_graphs(data):
+    # Create a color palette that is consistent across graphs
+    unique_sequences = sorted(set(seq for seq in data))
+    palette = sns.color_palette("hsv", len(unique_sequences))
+    color_dict = {seq: color for seq, color in zip(unique_sequences, palette)}
+
+    for model_name in models.keys():
+        fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(16, 6), sharey=True)
+        for prediction_val in [0, 1]:
+            ax = ax1 if prediction_val == 0 else ax2
+            filtered_data = {seq: values[model_name] for seq, values in data.items() if values[model_name][0] == prediction_val}
+            # Sorting sequences based on confidence, descending
+            sorted_sequences = sorted(filtered_data.items(), key=lambda x: x[1][1], reverse=True)
+            sequences = [x[0] for x in sorted_sequences]
+            conf_values = [x[1][1] for x in sorted_sequences]
+            colors = [color_dict[seq] for seq in sequences]
+            sns.barplot(x=sequences, y=conf_values, palette=colors, ax=ax)
+            ax.set_title(f'Confidence Scores for {model_name.capitalize()} (Prediction {prediction_val})')
+            ax.set_xlabel('Sequences')
+            ax.set_ylabel('Confidence')
+            ax.tick_params(axis='x', rotation=45)  # Rotate x labels for better visibility
+
+        st.pyplot(fig)