add a show examples action

Author: gabriben

src/wraval/actions/action_examples.py

@@ -0,0 +1,65 @@
+#
+# // Copyright Amazon.com, Inc. or its affiliates. All Rights Reserved.
+# // SPDX-License-Identifier: Apache-2.0
+#
+from dynaconf import Dynaconf
+from wraval.actions.data_utils import load_latest_dataset
+import pandas as pd
+from typing import Optional
+
+
+def show_examples(settings: Dynaconf, tone: Optional[str] = None, n_examples: int = 3) -> None:
+    """
+    Load the latest dataset and display examples grouped by tone and model.
+    
+    Args:
+        settings: Dynaconf settings object with data_dir setting
+        tone: Optional tone to filter by
+        n_examples: Number of examples to show per tone-model combination
+    """
+    try:
+        # Use settings.data_dir which could be either local path or S3 URI
+        data_location = settings.data_dir
+        print(f"Loading data from: {data_location}")
+        d = load_latest_dataset(data_location)
+        
+        if tone and tone != "all":
+            if tone not in d['tone'].unique():
+                print(f"Error: Tone '{tone}' not found in dataset.")
+                print(f"Available tones: {', '.join(d['tone'].unique())}")
+                return
+            d = d[d['tone'] == tone]
+            print(f"\nExamples for Tone: {tone}")
+        else:
+            print("\nExamples by Tone and Model:")
+        
+        # Get unique combinations of tone and inference_model
+        combinations = d[['tone', 'inference_model']].drop_duplicates()
+        
+        for _, (tone, model) in combinations.iterrows():
+            print("\n" + "=" * 80)
+            print(f"Tone: {tone} | Model: {model}")
+            print("=" * 80)
+            
+            # Get examples for this tone-model combination
+            examples = d[(d['tone'] == tone) & (d['inference_model'] == model)]
+            
+            # Sample n_examples if we have more
+            if len(examples) > n_examples:
+                examples = examples.sample(n=n_examples, random_state=42)
+            
+            # Display each example
+            for idx, row in examples.iterrows():
+                print(f"\nExample {idx + 1}:")
+                print(f"Original: {row['synthetic_data']}")
+                print(f"Rewrite:  {row['rewrite']}")
+                print(f"Score:    {row['overall_score']:.2f}")
+                print("-" * 40)
+            
+    except FileNotFoundError as e:
+        print(f"Error: {e}")
+        print("Please generate and judge data first.")
+    except KeyError as e:
+        print(f"Error: Missing required column {e}. Please ensure the dataset has been properly judged.")
+    except Exception as e:
+        print(f"Unexpected error: {e}") 

src/wraval/main.py

@@ -12,6 +12,7 @@
 from wraval.actions.aws_utils import get_current_aws_account_id
 from wraval.actions.action_results import show_results
 from wraval.actions.action_deploy import deploy
+from wraval.actions.action_examples import show_examples
 import os
 
 
@@ -57,6 +58,7 @@ def parse_args() -> argparse.Namespace:
             "human_judge_upload",
             "human_judge_parsing",
             "show_results",
+            "show_examples",
             "deploy"
         ],
         help="Action to perform (generate data or run inference)",
@@ -124,6 +126,9 @@ def handle_judge(args, settings):
 def handle_show_results(args, settings):
     show_results(settings, args.type)
 
+def handle_show_examples(args, settings):
+    show_examples(settings, args.type)
+
 def handle_deploy(settings):
     deploy(settings)
 
@@ -140,6 +145,8 @@ def main():
             handle_judge(args, settings)
         case "show_results":
             handle_show_results(args, settings)
+        case "show_examples":
+            handle_show_examples(args, settings)
         case "deploy":
             handle_deploy(settings)
         case _: