Doc update for using INC for GPTQ INT4 (#3485)

* Doc update for using INC for GPTQ INT4

* update per comment

* remove instruction for deprecated example

* update DeepSpeed parts

* cherry-pick #3486

* correct BKC for accuracy testing with deepspeed

---------

Co-authored-by: Chunyuan WU <[email protected]>

Author: ZailiWang

docker/Dockerfile.prebuilt

@@ -40,7 +40,8 @@ ARG TORCHCCL_VERSION=2.6.0
 ARG PYTORCH_VERSION=2.6.0
 ARG TORCHAUDIO_VERSION=2.6.0
 ARG TORCHVISION_VERSION=0.21.0
-RUN . ./venv/bin/activate && \
+RUN python -m venv venv && \
+    . ./venv/bin/activate && \
     python -m pip --no-cache-dir install --upgrade \
     pip \
     setuptools \

examples/cpu/inference/python/llm-modeling/README.md

@@ -72,7 +72,7 @@ conda install gperftools -y
 python -m pip install transformers==4.38.1 accelerate
 
 # Set the environment variables for performance on Xeon
-export LD_PRELOAD=$(bash ../../../llm/tools/get_libstdcpp_lib.sh):${CONDA_PREFIX}/lib/libiomp.so:${CONDA_PREFIX}/lib/libtcmalloc.so:${LD_PRELOAD}
+export LD_PRELOAD=$(bash ../../../llm/tools/get_libstdcpp_lib.sh):${CONDA_PREFIX}/lib/libiomp5.so:${CONDA_PREFIX}/lib/libtcmalloc.so:${LD_PRELOAD}
 export KMP_BLOCKTIME=1
 export KMP_TPAUSE=0
 export KMP_FORKJOIN_BARRIER_PATTERN=dist,dist