fix

Dockerfile

@@ -13,4 +13,4 @@ COPY --chown=user . /app
 
 RUN python -m spacy download en_core_web_sm
 
-CMD ["gunicorn", "--bind", "0.0.0.0:7860", "app:app"]
+CMD ["gunicorn", "--bind", "0.0.0.0:7860", "--timeout", "360", "--workers", "1", "--log-level", "debug", "app:app"]

create_app.py

@@ -19,6 +19,9 @@ MODEL_SESSION = None
 def load_models():
     global MODELS_LOADED, LONGFORMER_TOKENIZER, LONGFORMER_MODEL, QWEN_TOKENIZER, QWEN_MODEL
     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+    print(f"CUDA Available: {torch.cuda.is_available()}")  
+    print(f"CUDA Device Name: {torch.cuda.get_device_name(0)}")  
+    print(f"Free GPU Memory: {torch.cuda.memory_reserved(0)/1e9:.2f} GB")  
     print("DEIVCE=", device)
     print("WHY DONT PRINT")
     if not MODELS_LOADED:
@@ -33,7 +36,7 @@ def load_models():
         QWEN_TOKENIZER.pad_token_id = QWEN_TOKENIZER.eos_token_id
         print("QWEN TOKENIZER LOADED")
         try:
-            QWEN_MODEL = AutoModelForCausalLM.from_pretrained(model_name).half()
+            QWEN_MODEL = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float16, device_map="auto", trust_remote_code=True)
             QWEN_MODEL = QWEN_MODEL.to(device)
             print("QWEN MODEL LOADED")  # Typo fixed from your original code ("LOADED" vs. "LOADED")
         except Exception as e: