backup

app.py

@@ -134,14 +134,25 @@ def save_evaluation(post_id, model_a, model_b, verdict):
 
 
 def get_random_sample():
-    """Get a random sample from the dataset. Sampling with replacement is allowed."""
-    idx = random.randint(0, len(dataset) - 1)
-    sample = dataset[idx]
+    """Get a random sample by first selecting a post_id then picking random edits for that post."""
+    # First randomly select a post_id from valid posts
+    random_post_id = random.choice(list(valid_post_ids))
+
+    # Filter dataset for this post_id using batched processing
+    post_edits = dataset.filter(
+        lambda xs: [x == random_post_id for x in xs["post_id"]],
+        batched=True,
+        batch_size=256,
+    )
+
+    # Get matching photoexp entries for this post_id
+    matching_photoexp_entries = photoexp[photoexp.post_id == random_post_id]
 
-    # Filter photoexp to get entries with the same post_id
-    matching_photoexp_entries = photoexp[photoexp.post_id == sample["post_id"]]
+    # Randomly select one edit from the dataset
+    idx = random.randint(0, len(post_edits) - 1)
+    sample = post_edits[idx]
 
-    # Randomly select one entry from the matching entries
+    # Randomly select one entry from the matching photoexp entries
     if not matching_photoexp_entries.empty:
         random_photoexp_entry = matching_photoexp_entries.sample(n=1).iloc[0]
         additional_edited_image = random_photoexp_entry["edited_image"]
@@ -164,7 +175,8 @@ def get_random_sample():
         image_b = sample["edited_image"]
         model_b = sample["model"]
 
-    print(f"Selected sample {idx}: {sample['post_id']} - {sample['model']}")
+    print(f"Selected post_id: {random_post_id}")
+    print(f"Selected edit from model: {sample['model']}")
 
     return {
         "post_id": sample["post_id"],