debugging

app.py

@@ -23,6 +23,9 @@ def chat_with_model(messages):
         yield messages + [{"role": "assistant", "content": "⚠️ No model loaded."}]
         return
 
+    print("EOS token ID:", current_tokenizer.eos_token_id)
+    print(current_model.generation_config)
+
     current_model.to("cuda").half()
 
     prompt = format_prompt(messages)