diff --git a/examples/awq/fp8_dynamic_llama_example.py b/examples/awq/fp8_dynamic_llama_example.py index 02018e56ee..1c7a6fb15e 100644 --- a/examples/awq/fp8_dynamic_llama_example.py +++ b/examples/awq/fp8_dynamic_llama_example.py @@ -76,6 +76,6 @@ def tokenize(sample): print("==========================================\n\n") # Save to disk compressed. -SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + "-awq-asym" +SAVE_DIR = MODEL_ID.rstrip("/").split("/")[-1] + "-awq-fp8-dynamic" model.save_pretrained(SAVE_DIR, save_compressed=True) tokenizer.save_pretrained(SAVE_DIR)