初始化项目，由ModelHub XC社区提供模型

Model: RedHatAI/QwQ-32B-Preview-quantized.w8a8
Source: Original Platform

quant_w8a8.sh

@@ -0,0 +1,22 @@
+#!/bin/bash
+
+    # for DAMPENING_FRAC 0.01 0.1;
+    # do
+    #     for OBSERVER minmax mse;
+    #     do
+    #         for ACTORDER False group;
+    #         do
+
+# export DAMPENING_FRAC=0.01
+# export OBSERVER="minmax"
+
+export CUDA_VISIBLE_DEVICES=${1}
+export MDL=${2}
+export OBSERVER=${3} # minmax mse
+export DAMPENING_FRAC=${4} # 0.01 0.1
+
+for CALIB_SIZE in 128 512 1024;
+do
+    python w8a8.py --model_path ${MDL} --quant_path "output_dir_w8a8/${MDL}/calib${CALIB_SIZE}_eosFalse_damp${DAMPENING_FRAC}_obs${OBSERVER}" --calib_size ${CALIB_SIZE} --dampening_frac ${DAMPENING_FRAC} --observer ${OBSERVER}
+done
+