Update app.py

app.py

@@ -17,7 +17,7 @@ import time
 import logging
 
 # Setup logging
-logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s")
+logging.basicConfig(level=logging.INFO, format="%(asctime)s - %(levelname)s - %(message)s", filename="/home/user/clinical_oversight_analyzer.log")
 logger = logging.getLogger(__name__)
 
 # Persistent directory
@@ -34,10 +34,12 @@ for directory in [model_cache_dir, tool_cache_dir, file_cache_dir, report_dir, v
     os.makedirs(directory, exist_ok=True)
 
 os.environ["HF_HOME"] = model_cache_dir
-os.environ["TRANSFORMERS_CACHE"] = model_cache_dir
 os.environ["VLLM_CACHE_DIR"] = vllm_cache_dir
 os.environ["TOKENIZERS_PARALLELISM"] = "false"
 os.environ["CUDA_LAUNCH_BLOCKING"] = "1"
+# Remove TRANSFORMERS_CACHE to suppress warning
+if "TRANSFORMERS_CACHE" in os.environ:
+    del os.environ["TRANSFORMERS_CACHE"]
 
 current_dir = os.path.dirname(os.path.abspath(__file__))
 src_path = os.path.abspath(os.path.join(current_dir, "src"))
@@ -195,10 +197,9 @@ def init_agent():
         tool_files_dict={"new_tool": target_tool_path},
         force_finish=True,
         enable_checker=True,
-        step_rag_num=1,  # Reduced for speed
+        step_rag_num=1,
         seed=100,
         additional_default_tools=[],
-        num_engine_threads=1,  # Limit VLLM threads for stability
     )
     agent.init_model()
     log_system_usage("After Load")
@@ -230,8 +231,8 @@ def process_batch(agent, chunks: List[str], cache_path: str, prompt_template: st
             message=prompt,
             history=[],
             temperature=0.2,
-            max_new_tokens=256,  # Reduced for speed
-            max_token=1024,      # Reduced for speed
+            max_new_tokens=256,
+            max_token=1024,
             call_agent=False,
             conversation=[],
         ):