Upload folder using huggingface_hub

app.py

@@ -72,7 +72,8 @@ class ModelHandler:
                 self.model_path,
                 torch_dtype=torch.bfloat16 if torch.cuda.is_available() else torch.float32,
                 device_map="auto",
-                attn_implementation="flash_attention_2" if use_flash_attention else "eager",
+                # attn_implementation="flash_attention_2" if use_flash_attention else "sdpa",
+                attn_implementation="sdpa",
                 trust_remote_code=True
             )
             print("✅ Model loaded successfully!")