From 6935660c4ba442eff56932bec66157a8d33f0e7c Mon Sep 17 00:00:00 2001
From: weimingc <17592131+meenchen@users.noreply.github.com>
Date: Thu, 8 Jan 2026 20:16:56 +0000
Subject: [PATCH] minor fix, update autoquant example cmd

Signed-off-by: weimingc <17592131+meenchen@users.noreply.github.com>
---
 examples/llm_ptq/README.md        |  2 +-
 examples/llm_ptq/example_utils.py | 12 ++++++------
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/examples/llm_ptq/README.md b/examples/llm_ptq/README.md
index 23ab1ecf9..1ceb213c7 100755
--- a/examples/llm_ptq/README.md
+++ b/examples/llm_ptq/README.md
@@ -226,7 +226,7 @@ export HF_PATH=<the downloaded LLaMA checkpoint from the Hugging Face hub, or si
 # --auto_quantize_bits specifies the constraint for `AutoQuantize`
 # --quant specifies the formats to be searched for `AutoQuantize`
 # NOTE: auto_quantize_bits cannot be lower than the number of bits for the smallest quantization format in --quant
-scripts/huggingface_example.sh --type llama --model $HF_PATH --quant w4a8_awq,fp8 --auto_quantize_bits 4.8 --tp [1|2|4|8]  --calib_batch_size 4
+scripts/huggingface_example.sh --model $HF_PATH --quant w4a8_awq,fp8 --auto_quantize_bits 4.8 --calib_batch_size 4
 ```
 
 The above example perform `AutoQuantize` where the less quantization accuracy sensitive layers are quantized with `w4a8_awq` (specified by `--quant w4a8_awq`) and the more sensitive layers
diff --git a/examples/llm_ptq/example_utils.py b/examples/llm_ptq/example_utils.py
index 99c5cb95a..b4cf3b574 100755
--- a/examples/llm_ptq/example_utils.py
+++ b/examples/llm_ptq/example_utils.py
@@ -180,12 +180,12 @@ def build_quant_cfg(
         quant_cfg["quant_cfg"]["*image*"] = {"enable": False}
         quant_cfg["quant_cfg"]["*vision*"] = {"enable": False}
 
-        if model_type in ["qwen3moe", "qwen3next"] and qformat == "nvfp4":
-            # Disable the attention projection layers to retain accuracy
-            quant_cfg["quant_cfg"]["model*.*attn*in_proj*"] = {"enable": False}
-            quant_cfg["quant_cfg"]["model*.*attn*q_proj*"] = {"enable": False}
-            quant_cfg["quant_cfg"]["model*.*attn*k_proj*"] = {"enable": False}
-            quant_cfg["quant_cfg"]["model*.*attn*v_proj*"] = {"enable": False}
+    if model_type in ["qwen3moe", "qwen3next"] and qformat == "nvfp4":
+        # Disable the attention projection layers to retain accuracy
+        quant_cfg["quant_cfg"]["model*.*attn*in_proj*"] = {"enable": False}
+        quant_cfg["quant_cfg"]["model*.*attn*q_proj*"] = {"enable": False}
+        quant_cfg["quant_cfg"]["model*.*attn*k_proj*"] = {"enable": False}
+        quant_cfg["quant_cfg"]["model*.*attn*v_proj*"] = {"enable": False}
 
     return quant_cfg