hard问题分类

layer2/PGEE/code/classify_muti.py

@@ -12,7 +12,8 @@ API_KEY="sk-oYh3Xrhg8oDY2gW02c966f31C84449Ad86F9Cd9dF6E64a8d"
 BASE_URL="https://vip.apiyi.com/v1"
 MODEL_DEEPSEEK_V3 = "deepseek-chat"
 CATEGORIES = ['Atomic Structure and Interatomic Bonding', 'The Structure of Solids', 'Imperfections in Solids', 'Mechanical Properties of Metals','Dislocations and Strengthening Mechanisms','Failure','Phase Transformations: Development of Microstructure and Alteration of Mechanical Properties','Applications and Processing of Materials','Corrosion and Degradation of Materials','Functional Properties of Materials','Unknown']
-
+FILE_PATH = '/home/ubuntu/50T/fsy/A/MatBench/layer2/PGEE/code/stepz_final_choice_questions_filtered_only_hard.json'
+OUTPUT_PATH='/home/ubuntu/50T/fsy/A/MatBench/layer2/PGEE/code/stepz_classified_only_hard.json'
 # Thread-local storage for OpenAI clients
 local = threading.local()
 
@@ -133,8 +134,7 @@ def process_item(args):
 
 def main():
     # 加载数据
-    file_path = '/home/ubuntu/50T/fsy/MatBench/layer2/PGEE/code/stepz_final_choice_questions_filtered.json'
+    data = load_data(FILE_PATH)
-    data = load_data(file_path)
     data_length = len(data)
     
     results = []
@@ -143,7 +143,7 @@ def main():
     args_list = [(i, data_length, item) for i, item in enumerate(data)]
     
     # 设定线程数，根据实际API限制和服务器性能调整
-    num_threads = 10  # 根据需要调整线程数
+    num_threads = 20  # 根据需要调整线程数
     
     print(f"Starting classification with {num_threads} threads...")
     
@@ -154,7 +154,7 @@ def main():
         results = futures
     
     # 保存最终结果
-    with open('/home/ubuntu/50T/fsy/MatBench/layer2/PGEE/code/stepz_classified.json', 'w', encoding='utf-8') as f:
+    with open(OUTPUT_PATH, 'w', encoding='utf-8') as f:
         json.dump(results, f, ensure_ascii=False, indent=4)
     
     # 分析结果
@@ -163,7 +163,7 @@ def main():
     print("\nFinal distribution of questions by category:")
     print(category_counts_final)
     
-    print("\nTask completed. Results saved to '/home/ubuntu/50T/fsy/MatBench/layer2/PGEE/code/stepz_classified.json'")
+    print(f"\nTask completed. Results saved to {OUTPUT_PATH}")
 
 if __name__ == "__main__":
     main()