Use built image opea/vllm-gaudi:1.4 for AgentQnA, CodeGen, CodeTrans, DocSum and VisualQnA. (opea-project#2277)

Signed-off-by: ZePan110 <[email protected]>

Author: ZePan110

AgentQnA/docker_compose/intel/hpu/gaudi/compose.yaml

@@ -128,7 +128,7 @@ services:
       OPENAI_API_BASE_URLS: ${SUPERVISOR_AGENT_ENDPOINT}
       ENABLE_OLLAMA_API: False
   vllm-service:
-    image: ${REGISTRY:-opea}/vllm-gaudi:${TAG:-latest}
+    image: opea/vllm-gaudi:1.4
     container_name: vllm-gaudi-server
     ports:
       - "8086:8000"

AgentQnA/docker_image_build/build.yaml

@@ -17,12 +17,6 @@ services:
       dockerfile: ./docker/Dockerfile
     extends: agent
     image: ${REGISTRY:-opea}/agent-ui:${TAG:-latest}
-  vllm-gaudi:
-    build:
-      context: vllm-fork
-      dockerfile: Dockerfile.hpu
-    extends: agent
-    image: ${REGISTRY:-opea}/vllm-gaudi:${TAG:-latest}
   vllm-rocm:
     build:
       context: GenAIComps

AgentQnA/tests/step1_build_images.sh

@@ -40,12 +40,8 @@ function build_agent_docker_image_gaudi_vllm() {
     cd $WORKDIR/GenAIExamples/AgentQnA/docker_image_build/
     get_genai_comps
 
-    git clone https://github.com/HabanaAI/vllm-fork.git && cd vllm-fork
-    VLLM_FORK_VER=v0.6.6.post1+Gaudi-1.20.0
-    git checkout ${VLLM_FORK_VER} &> /dev/null && cd ../
-
     echo "Build agent image with --no-cache..."
-    service_list="agent agent-ui vllm-gaudi"
+    service_list="agent agent-ui"
     docker compose -f build.yaml build ${service_list} --no-cache
 }
 
@@ -83,6 +79,7 @@ function main() {
         "rocm_vllm")
             echo "==================== Build agent docker image for ROCm VLLM ===================="
             build_agent_docker_image_rocm_vllm
+            docker image ls | grep vllm
             ;;
         "gaudi_vllm")
             echo "==================== Build agent docker image for Gaudi ===================="
@@ -97,8 +94,6 @@ function main() {
             exit 1
             ;;
     esac
-
-    docker image ls | grep vllm
 }
 
 main $1

CodeGen/docker_compose/intel/hpu/gaudi/compose.yaml

@@ -3,7 +3,7 @@
 
 services:
   vllm-service:
-    image: ${REGISTRY:-opea}/vllm-gaudi:${TAG:-latest}
+    image: opea/vllm-gaudi:1.4
     container_name: vllm-gaudi-server
     ports:
       - "8028:80"

CodeGen/docker_image_build/build.yaml

@@ -78,12 +78,6 @@ services:
       dockerfile: comps/third_parties/vllm/src/Dockerfile.amd_gpu
     extends: codegen
     image: ${REGISTRY:-opea}/vllm-rocm:${TAG:-latest}
-  vllm-gaudi:
-    build:
-      context: vllm-fork
-      dockerfile: Dockerfile.hpu
-    extends: codegen
-    image: ${REGISTRY:-opea}/vllm-gaudi:${TAG:-latest}
   dataprep:
     build:
       context: GenAIComps

CodeGen/tests/test_compose_on_gaudi.sh

@@ -25,14 +25,8 @@ function build_docker_images() {
     docker build --no-cache -t ${REGISTRY}/comps-base:${TAG} --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f Dockerfile .
     popd && sleep 1s
 
-    # Download Gaudi vllm of latest tag
-    git clone https://github.com/HabanaAI/vllm-fork.git && cd vllm-fork
-    VLLM_FORK_VER=v0.6.6.post1+Gaudi-1.20.0
-    echo "Check out vLLM tag ${VLLM_FORK_VER}"
-    git checkout ${VLLM_FORK_VER} &> /dev/null && cd ../
-
     echo "Build all the images with --no-cache, check docker_image_build.log for details..."
-    service_list="codegen codegen-gradio-ui llm-textgen vllm-gaudi dataprep retriever embedding"
+    service_list="codegen codegen-gradio-ui llm-textgen dataprep retriever embedding"
     docker compose -f build.yaml build ${service_list} --no-cache > ${LOG_PATH}/docker_image_build.log
 
     docker images && sleep 1s

CodeTrans/docker_compose/intel/hpu/gaudi/compose.yaml

@@ -3,7 +3,7 @@
 
 services:
   vllm-service:
-    image: ${REGISTRY:-opea}/vllm-gaudi:${TAG:-latest}
+    image: opea/vllm-gaudi:1.4
     container_name: codetrans-gaudi-vllm-service
     ports:
       - "8008:80"

CodeTrans/docker_image_build/build.yaml

@@ -25,12 +25,6 @@ services:
       dockerfile: comps/llms/src/text-generation/Dockerfile
     extends: codetrans
     image: ${REGISTRY:-opea}/llm-textgen:${TAG:-latest}
-  vllm-gaudi:
-    build:
-      context: vllm-fork
-      dockerfile: Dockerfile.hpu
-    extends: codetrans
-    image: ${REGISTRY:-opea}/vllm-gaudi:${TAG:-latest}
   nginx:
     build:
       context: GenAIComps

CodeTrans/tests/test_compose_on_gaudi.sh

@@ -25,12 +25,8 @@ function build_docker_images() {
     docker build --no-cache -t ${REGISTRY}/comps-base:${TAG} --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f Dockerfile .
     popd && sleep 1s
 
-    git clone https://github.com/HabanaAI/vllm-fork.git && cd vllm-fork
-    VLLM_FORK_VER=v0.6.6.post1+Gaudi-1.20.0
-    git checkout ${VLLM_FORK_VER} &> /dev/null && cd ../
-
     echo "Build all the images with --no-cache, check docker_image_build.log for details..."
-    service_list="codetrans codetrans-ui llm-textgen vllm-gaudi nginx"
+    service_list="codetrans codetrans-ui llm-textgen nginx"
     docker compose -f build.yaml build ${service_list} --no-cache > ${LOG_PATH}/docker_image_build.log
 
     docker images && sleep 1s

DocSum/docker_compose/intel/hpu/gaudi/compose.yaml

@@ -3,7 +3,7 @@
 
 services:
   vllm-service:
-    image: ${REGISTRY:-opea}/vllm-gaudi:${TAG:-latest}
+    image: opea/vllm-gaudi:1.4
     container_name: docsum-gaudi-vllm-service
     ports:
       - ${LLM_ENDPOINT_PORT:-8008}:80