Using HF_HOME instead of CACHE to get token read in addition to models. (#2288)

Dockerfile

@@ -191,7 +191,7 @@ ENV PATH=/opt/conda/bin:$PATH \
     CONDA_PREFIX=/opt/conda
 
 # Text Generation Inference base env
-ENV HUGGINGFACE_HUB_CACHE=/data \
+ENV HF_HOME=/data \
     HF_HUB_ENABLE_HF_TRANSFER=1 \
     PORT=80
 

Dockerfile_amd

@@ -167,7 +167,7 @@ RUN python setup.py build
 FROM base AS base-copy
 
 # Text Generation Inference base env
-ENV HUGGINGFACE_HUB_CACHE=/data \
+ENV HF_HOME=/data \
     HF_HUB_ENABLE_HF_TRANSFER=1 \
     PORT=80
 

Dockerfile_intel

@@ -57,7 +57,7 @@ RUN wget -O- https://apt.repos.intel.com/intel-gpg-keys/GPG-PUB-KEY-INTEL-SW-PRO
 RUN apt-get update && apt install -y intel-basekit xpu-smi cmake python3-dev ninja-build pciutils
 
 # Text Generation Inference base env
-ENV HUGGINGFACE_HUB_CACHE=/data \
+ENV HF_HOME=/data \
     HF_HUB_ENABLE_HF_TRANSFER=1 \
     PORT=80