Update app.py

app.py

@@ -18,8 +18,8 @@ import torch
 # filename = "simplescaling_s1-32B-Q4_K_S.gguf"
 # model_name = "simplescaling/s1.1-32B"
 # model_name = "unsloth/Llama-4-Scout-17B-16E-Instruct-GGUF"
-# model_name = "unsloth/Llama-4-Scout-17B-16E-Instruct-unsloth-bnb-4bit"
-model_name = "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+model_name = "unsloth/Llama-4-Scout-17B-16E-Instruct-unsloth-bnb-4bit"
+# model_name = "meta-llama/Llama-4-Scout-17B-16E-Instruct"
 filename = "Llama-4-Scout-17B-16E-Instruct-UD-IQ2_XXS.gguf"
 torch_dtype = torch.bfloat16 # could be torch.float16 or torch.bfloat16 torch.float32 too
 cache_dir = "/data"
@@ -35,7 +35,7 @@ cache_dir = "/data"
 model = Llama4ForConditionalGeneration.from_pretrained(
     model_name,
     # default is eager attention
-    attn_implementation="flex_attention",
+    # attn_implementation="flex_attention",
     # gguf_file=filename,
     cache_dir = cache_dir,
     torch_dtype=torch_dtype,