removes flash attention from app

app.py

@@ -119,7 +119,7 @@ def load_model():
             "torch_dtype": torch.bfloat16 if DEVICE == "cuda" else torch.float32,  # Use float16 on GPU, float32 on CPU
             "trust_remote_code": True,
             "low_cpu_mem_usage": True,
-            "attn_implementation": "flash_attention_2" if DEVICE == "cuda" else "eager"
+        #   "attn_implementation": "flash_attention_2" if DEVICE == "cuda" else "eager"
         }
         
         logger.info(f"Model loading parameters: {model_kwargs}")