Update README.md

README.md

@@ -34,7 +34,7 @@ import torch
 
 model = AutoModelForCausalLM.from_pretrained(
     "TevunahAi/granite-8b-code-instruct-4k-FP8",
-    torch_dtype=torch.float8_e4m3fn,  # FP8 dtype
+    torch_dtype=torch.bfloat16,
     device_map="auto",
     low_cpu_mem_usage=True,
 )