Update app.py

app.py

@@ -35,4 +35,48 @@ with gr.Blocks() as demo:
     
     run_button.click(fn=evaluate_prompt, inputs=[model_name, prompt_text], outputs=output)
 
-demo.launch()
+demo.launch()
+import pandas as pd
+
+def batch_evaluate(file, model_name):
+    df = pd.read_csv(file.name)  # CSV must have a column called 'prompt'
+    results = []
+    for idx, row in df.iterrows():
+        prompt_text = row['prompt']
+        response = openai.ChatCompletion.create(
+            model=model_name,
+            messages=[{"role":"user","content":prompt_text}],
+            temperature=0.5
+        )
+        results.append({
+            'prompt': prompt_text,
+            'model': model_name,
+            'response': response['choices'][0]['message']['content']
+        })
+    return pd.DataFrame(results)
+
+with gr.Blocks() as demo:
+    csv_input = gr.File(label="Upload CSV with 'prompt' column")
+    model_input = gr.Textbox(label="Model Name")
+    run_button = gr.Button("Run Batch Eval")
+    output_table = gr.Dataframe(headers=["prompt", "model", "response"])
+    
+    run_button.click(batch_evaluate, inputs=[csv_input, model_input], outputs=output_table)
+
+demo.launch()
+import io
+
+def results_to_csv(df):
+    buffer = io.StringIO()
+    df.to_csv(buffer, index=False)
+    buffer.seek(0)
+    return buffer
+
+download_btn = gr.Button("Download CSV")
+download_file = gr.File()
+
+download_btn.click(results_to_csv, inputs=output_table, outputs=download_file)
+def metrics(df):
+    df['length'] = df['response'].apply(len)
+    summary = df.groupby('model').agg(avg_length=('length','mean')).reset_index()
+    return summary