Update app.py

app.py

@@ -8,7 +8,7 @@ from datasets import load_dataset
 import accelerate
 
 # 환경 변수에서 토큰 가져오기
-hf_token = os.environ.get("HF_TOKEN", None)
+hf_token = os.environ.get("HF_TOKEN", "Accounting")
 
 # Hugging Face 로그인
 if hf_token:
@@ -40,13 +40,12 @@ def evaluate_model(question, choices):
     answer = tokenizer.decode(outputs[0][-1:], skip_special_tokens=True).strip()
     return answer
 
-def run_kmmlu_test(subject):
-    subject_df = df[df['subject'] == subject]
+def run_kmmlu_test():
     correct = 0
-    total = len(subject_df)
+    total = len(df)
     
     results = []
-    for _, row in subject_df.iterrows():
+    for _, row in df.iterrows():
         question = row['question']
         choices = [row['A'], row['B'], row['C'], row['D']]
         correct_answer = row['answer']
@@ -60,18 +59,17 @@ def run_kmmlu_test(subject):
         results.append(f"질문: {question}\n모델 답변: {model_answer}\n정답: {correct_answer}\n정확도: {'맞음' if is_correct else '틀림'}\n")
     
     accuracy = correct / total
-    summary = f"주제: {subject}\n정확도: {accuracy:.2%} ({correct}/{total})\n\n"
+    summary = f"전체 테스트 결과\n정확도: {accuracy:.2%} ({correct}/{total})\n\n"
     return summary + "\n".join(results)
 
-subjects=df['subject'].unique().tolist()
 
 iface = gr.Interface(
     fn=run_kmmlu_test,
-    #inputs="Accounting",
-    inputs=gr.Dropdown(choices=subjects, label="주제 선택"),
+    inputs=None,
+    #inputs=gr.Dropdown(choices=subjects, label="주제 선택"),
     outputs="text",
     title="Llama 3를 이용한 KMMLU 테스트",
-    description="선택한 주제에 대해 KMMLU 테스트를 실행합니다."
+    description="Accounting 영역에 대한 KMMLU 테스트 수행"
 )
 
 iface.launch()