Update scripts for official TGI Gaudi docker (#11)

* use official docker image Signed-off-by: lvliang-intel <liang1.lv@intel.com> * update scripts Signed-off-by: lvliang-intel <liang1.lv@intel.com> --------- Signed-off-by: lvliang-intel <liang1.lv@intel.com>
opea-project · Mar 23, 2024 · 997df1d · 997df1d
1 parent c84b2cd
commit 997df1d
Show file tree

Hide file tree

Showing 7 changed files with 19 additions and 11 deletions.
diff --git a/ChatQnA/serving/tgi_gaudi/build_docker.sh b/ChatQnA/serving/tgi_gaudi/build_docker.sh
@@ -2,4 +2,4 @@
 
 git clone https://github.com/huggingface/tgi-gaudi.git
 cd ./tgi-gaudi/
-docker build -t tgi_gaudi . --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy
+docker build -t ghcr.io/huggingface/tgi-gaudi:1.2.1 . --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy
diff --git a/ChatQnA/serving/tgi_gaudi/launch_tgi_service.sh b/ChatQnA/serving/tgi_gaudi/launch_tgi_service.sh
@@ -27,9 +27,9 @@ volume=$PWD/data
 
 # Build the Docker run command based on the number of cards
 if [ "$num_cards" -eq 1 ]; then
-    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi --model-id $model_name"
+    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi:1.2.1 --model-id $model_name"
 else
-    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e PT_HPU_ENABLE_LAZY_COLLECTIVES=true -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi --model-id $model_name --sharded true --num-shard $num_cards"
+    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e PT_HPU_ENABLE_LAZY_COLLECTIVES=true -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi:1.2.1 --model-id $model_name --sharded true --num-shard $num_cards"
 fi
 
 # Execute the Docker run command

diff --git a/CodeGen/README.md b/CodeGen/README.md
@@ -4,9 +4,17 @@ Code generation is a noteworthy application of Large Language Model (LLM) techno
 # Environment Setup
 To use [🤗 text-generation-inference](https://github.com/huggingface/text-generation-inference) on Intel Gaudi2, please follow these steps:
 
-## Build TGI Gaudi Docker Image
+## Prepare Gaudi Image
+Getting started is straightforward with the official Docker container. Simply pull the image using:
+
+```bash
+docker pull ghcr.io/huggingface/tgi-gaudi:1.2.1
+```
+
+Alternatively, you can build the Docker image yourself with:
+
 ```bash
-bash ./tgi_gaudi/build_docker.sh
+bash ./serving/tgi_gaudi/build_docker.sh
 ```
 
 ## Launch TGI Gaudi Service

diff --git a/CodeGen/tgi_gaudi/build_docker.sh b/CodeGen/tgi_gaudi/build_docker.sh
@@ -16,4 +16,4 @@
 
 git clone https://github.com/huggingface/tgi-gaudi.git
 cd ./tgi-gaudi/
-docker build -t tgi_gaudi_codegen . --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy
+docker build -t ghcr.io/huggingface/tgi-gaudi:1.2.1 . --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy
diff --git a/CodeGen/tgi_gaudi/launch_tgi_service.sh b/CodeGen/tgi_gaudi/launch_tgi_service.sh
@@ -41,9 +41,9 @@ volume=$PWD/data
 
 # Build the Docker run command based on the number of cards
 if [ "$num_cards" -eq 1 ]; then
-    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy tgi_gaudi_codegen --model-id $model_name"
+    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi:1.2.1 --model-id $model_name"
 else
-    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e PT_HPU_ENABLE_LAZY_COLLECTIVES=true -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy tgi_gaudi_codegen --model-id $model_name --sharded true --num-shard $num_cards"
+    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e PT_HPU_ENABLE_LAZY_COLLECTIVES=true -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi:1.2.1 --model-id $model_name --sharded true --num-shard $num_cards"
 fi
 
 # Execute the Docker run command

diff --git a/DocSum/serving/tgi_gaudi/build_docker.sh b/DocSum/serving/tgi_gaudi/build_docker.sh
@@ -2,4 +2,4 @@
 
 git clone https://github.com/huggingface/tgi-gaudi.git
 cd ./tgi-gaudi/
-docker build -t tgi_gaudi_doc_summary . --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy
+docker build -t ghcr.io/huggingface/tgi-gaudi:1.2.1 . --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy
diff --git a/DocSum/serving/tgi_gaudi/launch_tgi_service.sh b/DocSum/serving/tgi_gaudi/launch_tgi_service.sh
@@ -27,9 +27,9 @@ volume=$PWD/data
 
 # Build the Docker run command based on the number of cards
 if [ "$num_cards" -eq 1 ]; then
-    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy tgi_gaudi_doc_summary --model-id $model_name"
+    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi:1.2.1 --model-id $model_name"
 else
-    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e PT_HPU_ENABLE_LAZY_COLLECTIVES=true -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy tgi_gaudi_doc_summary --model-id $model_name --sharded true --num-shard $num_cards"
+    docker_cmd="docker run -p $port_number:80 -v $volume:/data --runtime=habana -e PT_HPU_ENABLE_LAZY_COLLECTIVES=true -e HABANA_VISIBLE_DEVICES=all -e OMPI_MCA_btl_vader_single_copy_mechanism=none --cap-add=sys_nice --ipc=host -e HTTPS_PROXY=$https_proxy -e HTTP_PROXY=$https_proxy ghcr.io/huggingface/tgi-gaudi:1.2.1 --model-id $model_name --sharded true --num-shard $num_cards"
 fi
 
 # Execute the Docker run command