Update app.py

app.py

@@ -146,7 +146,7 @@ model_n = AutoModelForImageTextToText.from_pretrained(
     MODEL_ID_N,
     trust_remote_code=True,
     torch_dtype=torch.float16,
-    attn_implementation="flash_attention_2"
+    attn_implementation="eager" # else "flash_attention_2"
 ).to(device).eval()
 
 
@@ -155,7 +155,7 @@ MODEL_PATH_D = model_path_d_local
 processor_d = AutoProcessor.from_pretrained(MODEL_PATH_D, trust_remote_code=True)
 model_d = AutoModelForCausalLM.from_pretrained(
     MODEL_PATH_D,
-    attn_implementation="flash_attention_2",
+    attn_implementation="eager", # else "flash_attention_2"
     torch_dtype=torch.bfloat16,
     device_map="auto",
     trust_remote_code=True