Update app.py

app.py

@@ -252,14 +252,27 @@ def generate_bugged_code(description, code, chat_history, is_first_time):
     encoded_code = encode_text(input_for_model)
     combined_input = f"Description: {description} _BREAK_ Code: {encoded_code}"
 
-    inputs = tokenizer.encode(combined_input, return_tensors="pt")
+    inputs = tokenizer(
+        combined_input,
+        return_tensors="pt",
+        truncation=True,
+        max_length=512,
+    ).input_ids.to(device)
 
-    outputs = model.generate(
-        inputs,
-        max_length=1024,
-        num_beams=5,
-        early_stopping=True,
-    )
+    try:
+        print("Starting generation...")
+        with torch.no_grad():
+            outputs = model.generate(
+                inputs,
+                max_new_tokens=256,
+                num_beams=1,
+                do_sample=False,
+                early_stopping=True,
+            )
+        print("Generation done.")
+    except Exception as e:
+        print("Generation error:", repr(e))
+        raise e
 
     bugged_code_escaped = tokenizer.decode(outputs[0], skip_special_tokens=True)
 
@@ -287,6 +300,7 @@ def generate_bugged_code(description, code, chat_history, is_first_time):
     return chat_history, gr.update(value=""), False
 
 
+
 def reset_interface():
     global current_code, bug_counter
     current_code = None
@@ -359,4 +373,4 @@ with gr.Blocks(title="Software-Fault Injection from NL") as demo:
     )
 
 print("Launching Gradio interface...")
-demo.launch()
+demo.queue(max_size=10, concurrency_count=1).launch()