Dtype

Xilinx · Jan 16, 2025 · 4642dce · 4642dce
1 parent 62b4284
commit 4642dce
Showing 1 changed file with 1 addition and 1 deletion.
diff --git a/src/brevitas_examples/llm/main.py b/src/brevitas_examples/llm/main.py
@@ -577,7 +577,7 @@ def quantize_llm(args):
             )
             model_config = TransformersModelConfig(
                 pretrained=args.model,
-                dtype="float16",
+                dtype=args.dtype,
                 use_chat_template=True,
                 model_parallel=True,
                 accelerator=accelerator,