fix qlora-finetune Dockerfile (#10379)

docker/llm/finetune/qlora/cpu/docker/Dockerfile

@@ -46,10 +46,8 @@ RUN mkdir -p /bigdl/data && mkdir -p /bigdl/model && \
     rm -r BigDL && \
     chown -R mpiuser /bigdl
 
-# for docker directly run example
+# for standalone
 COPY ./start-qlora-finetuning-on-cpu.sh /bigdl/start-qlora-finetuning-on-cpu.sh
-# for k8s
-COPY ./bigdl-qlora-finetuing-entrypoint.sh /bigdl/bigdl-qlora-finetuing-entrypoint.sh
 
 USER mpiuser
 

docker/llm/finetune/qlora/cpu/docker/Dockerfile.k8s

@@ -65,8 +65,8 @@ RUN mkdir -p /bigdl/data && mkdir -p /bigdl/model && \
     rm -r BigDL && \
     chown -R mpiuser /bigdl
 
-# for standalone
-COPY ./start-qlora-finetuning-on-cpu.sh /bigdl/start-qlora-finetuning-on-cpu.sh
+# for k8s
+COPY ./bigdl-qlora-finetuing-entrypoint.sh /bigdl/bigdl-qlora-finetuing-entrypoint.sh
 
 USER mpiuser
 

docker/llm/finetune/qlora/cpu/docker/bigdl-qlora-finetuing-entrypoint.sh

@@ -20,7 +20,7 @@ then
     -f /home/mpiuser/hostfile \
     -iface eth0 \
     --bind-to socket \
-    -genv OMP_NUM_THREADS=48 \
+    -genv OMP_NUM_THREADS=$OMP_NUM_THREADS \
     -genv KMP_AFFINITY="granularity=fine,none" \
     -genv KMP_BLOCKTIME=1 \
     -genv TF_ENABLE_ONEDNN_OPTS=1 \

docker/llm/finetune/qlora/cpu/kubernetes/templates/bigdl-qlora-finetuning-job.yaml

@@ -43,6 +43,8 @@ spec:
                value: "{{ .Values.httpProxy }}"
              - name: https_proxy
                value: "{{ .Values.httpsProxy }}"
+             - name: OMP_NUM_THREADS
+               value: "{{ .Values.omp_num_threads}}"
              - name: LOCAL_POD_NAME
                valueFrom:
                  fieldRef:

docker/llm/finetune/qlora/cpu/kubernetes/values.yaml

@@ -6,5 +6,6 @@ nfsServerIp: your_nfs_server_ip
 nfsPath: a_nfs_shared_folder_path_on_the_server
 dataSubPath: alpaca_data_cleaned_archive.json # a subpath of the data file under nfs directory
 modelSubPath: Llama-2-7b-chat-hf # a subpath of the model file (dir) under nfs directory
+omp_num_threads: 48 # configure this value based on the number of CPU cores
 httpProxy: "your_http_proxy_like_http://xxx:xxxx_if_needed_else_empty"
 httpsProxy: "your_https_proxy_like_http://xxx:xxxx_if_needed_else_empty"