Update app.py

app.py

@@ -6,7 +6,8 @@ import random
 import re
 
 # Load model and tokenizer
-model_name = "rgb2gbr/GRPO_BioMedmcqa_Qwen2.5-0.5B"
+# model_name = "rgb2gbr/GRPO_BioMedmcqa_Qwen2.5-0.5B"
+model_name = "rgb2gbr/BioXP-0.5B-MedMCQA"
 model = AutoModelForCausalLM.from_pretrained(model_name)
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 
@@ -49,7 +50,7 @@ def extract_answer(prediction: str) -> tuple:
 
 def predict(question: str, option_a: str, option_b: str, option_c: str, option_d: str, 
            correct_option: int = None, explanation: str = None,
-           temperature: float = 0.6, top_p: float = 0.9, max_tokens: int = 10):
+           temperature: float = 0.6, top_p: float = 0.9, max_tokens: int = 20):
     # Format the prompt
     prompt = f"Question: {question}\n\nOptions:\nA. {option_a}\nB. {option_b}\nC. {option_c}\nD. {option_d}\n\nAnswer:"
     
@@ -120,7 +121,7 @@ with gr.Blocks(title="Medical MCQ Predictor") as demo:
                 max_tokens = gr.Slider(
                     minimum=10,
                     maximum=512,
-                    value=256,
+                    value=20,
                     step=32,
                     label="Max Tokens",
                     info="Maximum length of the generated response"