Remove debug logging and swap Model A/B positions

Changes:
- Removed all debug print statements from app.py and html_generator.py
- Swapped model positions: RWKV7 is now Model A, Qwen3 is now Model B
- Green now indicates RWKV7 performs better, Red indicates Qwen3 performs better

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

app.py

@@ -208,13 +208,6 @@ def run_evaluation(text: str, progress=gr.Progress()):
             text,
             bos_mode="add_newline_token"
         )
-        print(f"[DEBUG] Qwen evaluation complete")
-        print(f"[DEBUG] Qwen top5_predictions type: {type(result_qwen.get('top5_predictions'))}")
-        print(f"[DEBUG] Qwen top5_predictions length: {len(result_qwen.get('top5_predictions', []))}")
-        if result_qwen.get('top5_predictions'):
-            print(f"[DEBUG] Qwen first prediction sample: {result_qwen['top5_predictions'][0]}")
-        print(f"[DEBUG] Qwen tokenizer type: {type(result_qwen.get('tokenizer'))}")
-        print(f"[DEBUG] Qwen tokenizer: {result_qwen.get('tokenizer')}")
 
         # Step 2: Evaluate RWKV7 (using cached model)
         progress(0, desc="Evaluating with RWKV7...")
@@ -223,39 +216,26 @@ def run_evaluation(text: str, progress=gr.Progress()):
             _rwkv_tokenizer,
             text
         )
-        print(f"[DEBUG] RWKV evaluation complete")
-        print(f"[DEBUG] RWKV top5_predictions type: {type(result_rwkv.get('top5_predictions'))}")
-        print(f"[DEBUG] RWKV top5_predictions length: {len(result_rwkv.get('top5_predictions', []))}")
-        if result_rwkv.get('top5_predictions'):
-            print(f"[DEBUG] RWKV first prediction sample: {result_rwkv['top5_predictions'][0]}")
-        print(f"[DEBUG] RWKV tokenizer type: {type(result_rwkv.get('tokenizer'))}")
-        print(f"[DEBUG] RWKV tokenizer: {result_rwkv.get('tokenizer')}")
 
         # Step 3: Generate visualization
         progress(0, desc="Generating visualization...")
-        print(f"[DEBUG] Starting HTML generation...")
-        print(f"[DEBUG] Passing tokenizer_a: {result_qwen['tokenizer']}")
-        print(f"[DEBUG] Passing tokenizer_b: {result_rwkv['tokenizer']}")
         html = generate_comparison_html(
             text=text,
-            byte_losses_a=result_qwen["byte_wise_losses"],
-            byte_losses_b=result_rwkv["byte_wise_losses"],
-            model_a_name="Qwen3-1.7B-Base",
-            model_b_name="RWKV7-G1C-1.5B",
-            topk_predictions_a=result_qwen["top5_predictions"],
-            topk_predictions_b=result_rwkv["top5_predictions"],
-            tokenizer_a=result_qwen["tokenizer"],
-            tokenizer_b=result_rwkv["tokenizer"],
-            model_type_a="hf",
-            model_type_b="rwkv7"
+            byte_losses_a=result_rwkv["byte_wise_losses"],
+            byte_losses_b=result_qwen["byte_wise_losses"],
+            model_a_name="RWKV7-G1C-1.5B",
+            model_b_name="Qwen3-1.7B-Base",
+            topk_predictions_a=result_rwkv["top5_predictions"],
+            topk_predictions_b=result_qwen["top5_predictions"],
+            tokenizer_a=result_rwkv["tokenizer"],
+            tokenizer_b=result_qwen["tokenizer"],
+            model_type_a="rwkv7",
+            model_type_b="hf"
         )
-        print(f"[DEBUG] HTML generation complete")
 
         # Wrap HTML for iframe display
         wrapped_html = wrap_html_in_iframe(html)
 
-        print("Done!")
-
         return wrapped_html
 
     except torch.cuda.OutOfMemoryError:

visualization/html_generator.py

@@ -274,25 +274,6 @@ def generate_comparison_html(
     model_a_token_ranges = build_byte_to_token_map(text, tokenizer_a, model_type_a)
     model_b_token_ranges = build_byte_to_token_map(text, tokenizer_b, model_type_b)
 
-    print(f"[DEBUG HTML] tokenizer_a: {tokenizer_a}")
-    print(f"[DEBUG HTML] tokenizer_b: {tokenizer_b}")
-    print(f"[DEBUG HTML] model_type_a: {model_type_a}")
-    print(f"[DEBUG HTML] model_type_b: {model_type_b}")
-    print(f"[DEBUG HTML] model_a_token_ranges length: {len(model_a_token_ranges)}")
-    print(f"[DEBUG HTML] model_b_token_ranges length: {len(model_b_token_ranges)}")
-    if model_a_token_ranges:
-        print(f"[DEBUG HTML] model_a first token range: {model_a_token_ranges[0]}")
-    if model_b_token_ranges:
-        print(f"[DEBUG HTML] model_b first token range: {model_b_token_ranges[0]}")
-    print(f"[DEBUG HTML] topk_predictions_a type: {type(topk_predictions_a)}")
-    print(f"[DEBUG HTML] topk_predictions_b type: {type(topk_predictions_b)}")
-    if topk_predictions_a:
-        print(f"[DEBUG HTML] topk_predictions_a length: {len(topk_predictions_a)}")
-        print(f"[DEBUG HTML] topk_predictions_a[0]: {topk_predictions_a[0]}")
-    if topk_predictions_b:
-        print(f"[DEBUG HTML] topk_predictions_b length: {len(topk_predictions_b)}")
-        print(f"[DEBUG HTML] topk_predictions_b[0]: {topk_predictions_b[0]}")
-
     def get_tokens_for_range(byte_start, byte_end, token_list):
         result = []
         for idx, (t_start, t_end, t_str) in enumerate(token_list):
@@ -411,9 +392,6 @@ def generate_comparison_html(
             model_a_token_idx = find_token_for_byte(byte_start, model_a_token_ranges)
             if model_a_token_idx is not None and model_a_token_idx < len(topk_predictions_a):
                 pred = topk_predictions_a[model_a_token_idx]
-                if token_count == 0:  # Only print for first token
-                    print(f"[DEBUG] Processing token at byte {byte_start}, model_a_token_idx={model_a_token_idx}")
-                    print(f"[DEBUG] pred structure: {pred}")
                 try:
                     decoded_pred = [
                         pred[0],
@@ -423,36 +401,19 @@ def generate_comparison_html(
                     # Use base64 encoding to avoid escaping issues
                     import base64
                     topk_a_json = base64.b64encode(json.dumps(decoded_pred, ensure_ascii=False).encode('utf-8')).decode('ascii')
-                    if token_count == 0:
-                        print(f"[DEBUG] Successfully generated topk_a_json (base64)")
-                        print(f"[DEBUG] Original JSON length: {len(json.dumps(decoded_pred, ensure_ascii=False))}")
-                        print(f"[DEBUG] Base64 length: {len(topk_a_json)}")
                 except Exception as e:
-                    print(f"[DEBUG] Error generating topk_a_json at byte {byte_start}: {e}")
-                    print(f"[DEBUG] pred[2] type: {type(pred[2])}")
-                    if len(pred) > 2:
-                        print(f"[DEBUG] pred[2] content: {pred[2][:3]}")
+                    pass
         if topk_predictions_b is not None and model_b_token_ranges:
             model_b_token_idx = find_token_for_byte(byte_start, model_b_token_ranges)
             if model_b_token_idx is not None and model_b_token_idx < len(topk_predictions_b):
                 pred = topk_predictions_b[model_b_token_idx]
-                if token_count == 0:  # Only print for first token
-                    print(f"[DEBUG] Processing token at byte {byte_start}, model_b_token_idx={model_b_token_idx}")
-                    print(f"[DEBUG] pred structure: {pred}")
                 try:
                     decoded_pred = [pred[0], pred[1], [[tid, prob, decode_token(tid, tokenizer_b, model_type_b)] for tid, prob in pred[2]]]
                     # Use base64 encoding to avoid escaping issues
                     import base64
                     topk_b_json = base64.b64encode(json.dumps(decoded_pred, ensure_ascii=False).encode('utf-8')).decode('ascii')
-                    if token_count == 0:
-                        print(f"[DEBUG] Successfully generated topk_b_json (base64)")
-                        print(f"[DEBUG] Original JSON length: {len(json.dumps(decoded_pred, ensure_ascii=False))}")
-                        print(f"[DEBUG] Base64 length: {len(topk_b_json)}")
                 except Exception as e:
-                    print(f"[DEBUG] Error generating topk_b_json at byte {byte_start}: {e}")
-                    print(f"[DEBUG] pred[2] type: {type(pred[2])}")
-                    if len(pred) > 2:
-                        print(f"[DEBUG] pred[2] content: {pred[2][:3]}")
+                    pass
 
         token_count += 1