Refactor llm Docsum (#1101)
Signed-off-by: Xinyao Wang <xinyao.wang@intel.com>
This commit is contained in:
XinyaoWa
152
tests/llms/test_llms_doc-summarization_langchain_tgi.sh
Normal file
152
tests/llms/test_llms_doc-summarization_langchain_tgi.sh
Normal file
@@ -0,0 +1,152 @@
|
||||
#!/bin/bash
|
||||
# Copyright (C) 2024 Intel Corporation
|
||||
# SPDX-License-Identifier: Apache-2.0
|
||||
|
||||
set -xe
|
||||
|
||||
WORKPATH=$(dirname "$PWD")
|
||||
host_ip=$(hostname -I | awk '{print $1}')
|
||||
LOG_PATH="$WORKPATH/tests"
|
||||
|
||||
function build_docker_images() {
|
||||
cd $WORKPATH
|
||||
docker build --no-cache -t opea/llm-docsum:latest --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f comps/llms/src/doc-summarization/Dockerfile .
|
||||
if [ $? -ne 0 ]; then
|
||||
echo "opea/llm-docsum built fail"
|
||||
exit 1
|
||||
else
|
||||
echo "opea/llm-docsum built successful"
|
||||
fi
|
||||
}
|
||||
|
||||
function start_service() {
|
||||
export host_ip=${host_ip}
|
||||
export LLM_ENDPOINT_PORT=5072
|
||||
export DOCSUM_PORT=5073
|
||||
export HUGGINGFACEHUB_API_TOKEN=${HF_TOKEN}
|
||||
export LLM_ENDPOINT="http://${host_ip}:${LLM_ENDPOINT_PORT}"
|
||||
export LLM_MODEL_ID="Intel/neural-chat-7b-v3-3"
|
||||
export MAX_INPUT_TOKENS=2048
|
||||
export MAX_TOTAL_TOKENS=4096
|
||||
export DocSum_COMPONENT_NAME="OPEADocSum_TGI" # or "vllm"
|
||||
export LOGFLAG=True
|
||||
|
||||
cd $WORKPATH/comps/llms/deployment/docker_compose
|
||||
docker compose -f doc-summarization_tgi.yaml up -d > ${LOG_PATH}/start_services_with_compose.log
|
||||
|
||||
sleep 30s
|
||||
}
|
||||
|
||||
function validate_services() {
|
||||
local URL="$1"
|
||||
local EXPECTED_RESULT="$2"
|
||||
local SERVICE_NAME="$3"
|
||||
local DOCKER_NAME="$4"
|
||||
local INPUT_DATA="$5"
|
||||
|
||||
local HTTP_STATUS=$(curl -s -o /dev/null -w "%{http_code}" -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL")
|
||||
|
||||
echo "==========================================="
|
||||
|
||||
if [ "$HTTP_STATUS" -eq 200 ]; then
|
||||
echo "[ $SERVICE_NAME ] HTTP status is 200. Checking content..."
|
||||
|
||||
local CONTENT=$(curl -s -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL" | tee ${LOG_PATH}/${SERVICE_NAME}.log)
|
||||
|
||||
echo $CONTENT
|
||||
|
||||
if echo "$CONTENT" | grep -q "$EXPECTED_RESULT"; then
|
||||
echo "[ $SERVICE_NAME ] Content is as expected."
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] Content does not match the expected result"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] HTTP status is not 200. Received status was $HTTP_STATUS"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
sleep 1s
|
||||
}
|
||||
|
||||
function validate_microservices() {
|
||||
URL="http://${host_ip}:$DOCSUM_PORT/v1/docsum"
|
||||
|
||||
echo "Validate tgi..."
|
||||
validate_services \
|
||||
"${LLM_ENDPOINT}/generate" \
|
||||
"generated_text" \
|
||||
"tgi" \
|
||||
"tgi-server" \
|
||||
'{"inputs":"What is Deep Learning?","parameters":{"max_new_tokens":17, "do_sample": true}}'
|
||||
|
||||
echo "Validate stream=True..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query": "Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en"}'
|
||||
|
||||
echo "Validate stream=False..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "stream":false}'
|
||||
|
||||
echo "Validate Chinese mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"2024年9月26日，北京——今日，英特尔正式发布英特尔® 至强® 6性能核处理器（代号Granite Rapids），为AI、数据分析、科学计算等计算密集型业务提供卓越性能。", "max_tokens":32, "language":"zh", "stream":false}'
|
||||
|
||||
echo "Validate truncate mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "truncate", "chunk_size": 2000}'
|
||||
|
||||
echo "Validate map_reduce mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "map_reduce", "chunk_size": 2000, "stream":false}'
|
||||
|
||||
echo "Validate refine mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "refine", "chunk_size": 2000}'
|
||||
}
|
||||
|
||||
function stop_docker() {
|
||||
cd $WORKPATH/comps/llms/deployment/docker_compose
|
||||
docker compose -f doc-summarization_tgi.yaml down
|
||||
}
|
||||
|
||||
function main() {
|
||||
|
||||
stop_docker
|
||||
|
||||
build_docker_images
|
||||
start_service
|
||||
|
||||
validate_microservices
|
||||
|
||||
stop_docker
|
||||
echo y | docker system prune
|
||||
|
||||
}
|
||||
|
||||
main
|
||||
@@ -0,0 +1,152 @@
|
||||
#!/bin/bash
|
||||
# Copyright (C) 2024 Intel Corporation
|
||||
# SPDX-License-Identifier: Apache-2.0
|
||||
|
||||
set -xe
|
||||
|
||||
WORKPATH=$(dirname "$PWD")
|
||||
host_ip=$(hostname -I | awk '{print $1}')
|
||||
LOG_PATH="$WORKPATH/tests"
|
||||
|
||||
function build_docker_images() {
|
||||
cd $WORKPATH
|
||||
docker build --no-cache -t opea/llm-docsum:latest --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f comps/llms/src/doc-summarization/Dockerfile .
|
||||
if [ $? -ne 0 ]; then
|
||||
echo "opea/llm-docsum built fail"
|
||||
exit 1
|
||||
else
|
||||
echo "opea/llm-docsum built successful"
|
||||
fi
|
||||
}
|
||||
|
||||
function start_service() {
|
||||
export host_ip=${host_ip}
|
||||
export LLM_ENDPOINT_PORT=5071
|
||||
export DOCSUM_PORT=5072
|
||||
export HUGGINGFACEHUB_API_TOKEN=${HF_TOKEN}
|
||||
export LLM_ENDPOINT="http://${host_ip}:${LLM_ENDPOINT_PORT}"
|
||||
export LLM_MODEL_ID="Intel/neural-chat-7b-v3-3"
|
||||
export MAX_INPUT_TOKENS=2048
|
||||
export MAX_TOTAL_TOKENS=4096
|
||||
export DocSum_COMPONENT_NAME="OPEADocSum_TGI" # or "vllm"
|
||||
export LOGFLAG=True
|
||||
|
||||
cd $WORKPATH/comps/llms/deployment/docker_compose
|
||||
docker compose -f doc-summarization_tgi_on_intel_hpu.yaml up -d > ${LOG_PATH}/start_services_with_compose.log
|
||||
|
||||
sleep 30s
|
||||
}
|
||||
|
||||
function validate_services() {
|
||||
local URL="$1"
|
||||
local EXPECTED_RESULT="$2"
|
||||
local SERVICE_NAME="$3"
|
||||
local DOCKER_NAME="$4"
|
||||
local INPUT_DATA="$5"
|
||||
|
||||
local HTTP_STATUS=$(curl -s -o /dev/null -w "%{http_code}" -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL")
|
||||
|
||||
echo "==========================================="
|
||||
|
||||
if [ "$HTTP_STATUS" -eq 200 ]; then
|
||||
echo "[ $SERVICE_NAME ] HTTP status is 200. Checking content..."
|
||||
|
||||
local CONTENT=$(curl -s -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL" | tee ${LOG_PATH}/${SERVICE_NAME}.log)
|
||||
|
||||
echo $CONTENT
|
||||
|
||||
if echo "$CONTENT" | grep -q "$EXPECTED_RESULT"; then
|
||||
echo "[ $SERVICE_NAME ] Content is as expected."
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] Content does not match the expected result"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] HTTP status is not 200. Received status was $HTTP_STATUS"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
sleep 1s
|
||||
}
|
||||
|
||||
function validate_microservices() {
|
||||
URL="http://${host_ip}:$DOCSUM_PORT/v1/docsum"
|
||||
|
||||
echo "Validate tgi..."
|
||||
validate_services \
|
||||
"${LLM_ENDPOINT}/generate" \
|
||||
"generated_text" \
|
||||
"tgi" \
|
||||
"tgi_gaudi_server" \
|
||||
'{"inputs":"What is Deep Learning?","parameters":{"max_new_tokens":17, "do_sample": true}}'
|
||||
|
||||
echo "Validate stream=True..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query": "Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en"}'
|
||||
|
||||
echo "Validate stream=False..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "stream":false}'
|
||||
|
||||
echo "Validate Chinese mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"2024年9月26日，北京——今日，英特尔正式发布英特尔® 至强® 6性能核处理器（代号Granite Rapids），为AI、数据分析、科学计算等计算密集型业务提供卓越性能。", "max_tokens":32, "language":"zh", "stream":false}'
|
||||
|
||||
echo "Validate truncate mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "truncate", "chunk_size": 2000}'
|
||||
|
||||
echo "Validate map_reduce mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "map_reduce", "chunk_size": 2000, "stream":false}'
|
||||
|
||||
echo "Validate refine mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "refine", "chunk_size": 2000}'
|
||||
}
|
||||
|
||||
function stop_docker() {
|
||||
cd $WORKPATH/comps/llms/deployment/docker_compose
|
||||
docker compose -f doc-summarization_tgi_on_intel_hpu.yaml down
|
||||
}
|
||||
|
||||
function main() {
|
||||
|
||||
stop_docker
|
||||
|
||||
build_docker_images
|
||||
start_service
|
||||
|
||||
validate_microservices
|
||||
|
||||
stop_docker
|
||||
echo y | docker system prune
|
||||
|
||||
}
|
||||
|
||||
main
|
||||
@@ -0,0 +1,165 @@
|
||||
#!/bin/bash
|
||||
# Copyright (C) 2024 Intel Corporation
|
||||
# SPDX-License-Identifier: Apache-2.0
|
||||
|
||||
set -xe
|
||||
|
||||
WORKPATH=$(dirname "$PWD")
|
||||
host_ip=$(hostname -I | awk '{print $1}')
|
||||
LOG_PATH="$WORKPATH/tests"
|
||||
|
||||
function build_docker_images() {
|
||||
cd $WORKPATH
|
||||
git clone https://github.com/HabanaAI/vllm-fork.git
|
||||
cd vllm-fork/
|
||||
git checkout 3c39626
|
||||
docker build --no-cache -f Dockerfile.hpu -t opea/vllm-gaudi:latest --shm-size=128g .
|
||||
if [ $? -ne 0 ]; then
|
||||
echo "opea/vllm-gaudi built fail"
|
||||
exit 1
|
||||
else
|
||||
echo "opea/vllm-gaudi built successful"
|
||||
fi
|
||||
|
||||
cd $WORKPATH
|
||||
docker build --no-cache -t opea/llm-docsum:latest --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f comps/llms/src/doc-summarization/Dockerfile .
|
||||
if [ $? -ne 0 ]; then
|
||||
echo "opea/llm-docsum built fail"
|
||||
exit 1
|
||||
else
|
||||
echo "opea/llm-docsum built successful"
|
||||
fi
|
||||
}
|
||||
|
||||
function start_service() {
|
||||
export host_ip=${host_ip}
|
||||
export LLM_ENDPOINT_PORT=5076
|
||||
export DOCSUM_PORT=5077
|
||||
export HUGGINGFACEHUB_API_TOKEN=${HF_TOKEN}
|
||||
export LLM_ENDPOINT="http://${host_ip}:${LLM_ENDPOINT_PORT}"
|
||||
export LLM_MODEL_ID="Intel/neural-chat-7b-v3-3"
|
||||
export MAX_INPUT_TOKENS=2048
|
||||
export MAX_TOTAL_TOKENS=4096
|
||||
export DocSum_COMPONENT_NAME="OPEADocSum_vLLM" # or "vllm"
|
||||
export VLLM_SKIP_WARMUP=true
|
||||
export LOGFLAG=True
|
||||
|
||||
cd $WORKPATH/comps/llms/deployment/docker_compose
|
||||
docker compose -f doc-summarization_vllm_on_intel_hpu.yaml up -d > ${LOG_PATH}/start_services_with_compose.log
|
||||
|
||||
sleep 30s
|
||||
}
|
||||
|
||||
function validate_services() {
|
||||
local URL="$1"
|
||||
local EXPECTED_RESULT="$2"
|
||||
local SERVICE_NAME="$3"
|
||||
local DOCKER_NAME="$4"
|
||||
local INPUT_DATA="$5"
|
||||
|
||||
local HTTP_STATUS=$(curl -s -o /dev/null -w "%{http_code}" -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL")
|
||||
|
||||
echo "==========================================="
|
||||
|
||||
if [ "$HTTP_STATUS" -eq 200 ]; then
|
||||
echo "[ $SERVICE_NAME ] HTTP status is 200. Checking content..."
|
||||
|
||||
local CONTENT=$(curl -s -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL" | tee ${LOG_PATH}/${SERVICE_NAME}.log)
|
||||
|
||||
echo $CONTENT
|
||||
|
||||
if echo "$CONTENT" | grep -q "$EXPECTED_RESULT"; then
|
||||
echo "[ $SERVICE_NAME ] Content is as expected."
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] Content does not match the expected result"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] HTTP status is not 200. Received status was $HTTP_STATUS"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
sleep 1s
|
||||
}
|
||||
|
||||
function validate_microservices() {
|
||||
URL="http://${host_ip}:$DOCSUM_PORT/v1/docsum"
|
||||
|
||||
echo "Validate vllm..."
|
||||
validate_services \
|
||||
"${LLM_ENDPOINT}/v1/completions" \
|
||||
"text" \
|
||||
"vllm" \
|
||||
"vllm-gaudi-server" \
|
||||
'{"model": "Intel/neural-chat-7b-v3-3", "prompt": "What is Deep Learning?", "max_tokens": 32, "temperature": 0}'
|
||||
|
||||
echo "Validate stream=True..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query": "Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en"}'
|
||||
|
||||
echo "Validate stream=False..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "stream":false}'
|
||||
|
||||
echo "Validate Chinese mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"2024年9月26日，北京——今日，英特尔正式发布英特尔® 至强® 6性能核处理器（代号Granite Rapids），为AI、数据分析、科学计算等计算密集型业务提供卓越性能。", "max_tokens":32, "language":"zh", "stream":false}'
|
||||
|
||||
echo "Validate truncate mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "truncate", "chunk_size": 2000}'
|
||||
|
||||
echo "Validate map_reduce mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "map_reduce", "chunk_size": 2000, "stream":false}'
|
||||
|
||||
echo "Validate refine mode..."
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"llm-docsum-server" \
|
||||
'{"query":"Text Embeddings Inference (TEI) is a toolkit for deploying and serving open source text embeddings and sequence classification models. TEI enables high-performance extraction for the most popular models, including FlagEmbedding, Ember, GTE and E5.", "max_tokens":32, "language":"en", "summary_type": "refine", "chunk_size": 2000}'
|
||||
}
|
||||
|
||||
function stop_docker() {
|
||||
cd $WORKPATH/comps/llms/deployment/docker_compose
|
||||
docker compose -f doc-summarization_vllm_on_intel_hpu.yaml down
|
||||
}
|
||||
|
||||
function main() {
|
||||
|
||||
stop_docker
|
||||
|
||||
build_docker_images
|
||||
start_service
|
||||
|
||||
validate_microservices
|
||||
|
||||
stop_docker
|
||||
echo y | docker system prune
|
||||
|
||||
}
|
||||
|
||||
main
|
||||
@@ -1,133 +0,0 @@
|
||||
#!/bin/bash
|
||||
# Copyright (C) 2024 Intel Corporation
|
||||
# SPDX-License-Identifier: Apache-2.0
|
||||
|
||||
set -xe
|
||||
|
||||
WORKPATH=$(dirname "$PWD")
|
||||
ip_address=$(hostname -I | awk '{print $1}')
|
||||
LOG_PATH="$WORKPATH/tests"
|
||||
|
||||
function build_docker_images() {
|
||||
cd $WORKPATH
|
||||
docker build --no-cache -t opea/llm-sum-tgi:comps --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f comps/llms/summarization/tgi/langchain/Dockerfile .
|
||||
if [ $? -ne 0 ]; then
|
||||
echo "opea/llm-textgen built fail"
|
||||
exit 1
|
||||
else
|
||||
echo "opea/llm-textgen built successful"
|
||||
fi
|
||||
}
|
||||
|
||||
function start_service() {
|
||||
tgi_endpoint_port=5075
|
||||
export LLM_MODEL_ID="Intel/neural-chat-7b-v3-3"
|
||||
export MAX_INPUT_TOKENS=2048
|
||||
export MAX_TOTAL_TOKENS=4096
|
||||
# Remember to set HF_TOKEN before invoking this test!
|
||||
export HF_TOKEN=${HF_TOKEN}
|
||||
docker run -d --name="test-comps-llm-sum-tgi-endpoint" -p $tgi_endpoint_port:80 -v ./data:/data -e http_proxy=$http_proxy -e https_proxy=$https_proxy --shm-size 1g ghcr.io/huggingface/text-generation-inference:1.4 --model-id ${LLM_MODEL_ID} --max-input-length ${MAX_INPUT_TOKENS} --max-total-tokens ${MAX_TOTAL_TOKENS}
|
||||
export TGI_LLM_ENDPOINT="http://${ip_address}:${tgi_endpoint_port}"
|
||||
|
||||
sum_port=5076
|
||||
docker run -d --name="test-comps-llm-sum-tgi-server" -p ${sum_port}:9000 --ipc=host -e http_proxy=$http_proxy -e https_proxy=$https_proxy -e TGI_LLM_ENDPOINT=$TGI_LLM_ENDPOINT -e LLM_MODEL_ID=$LLM_MODEL_ID -e MAX_INPUT_TOKENS=$MAX_INPUT_TOKENS -e MAX_TOTAL_TOKENS=$MAX_TOTAL_TOKENS -e HUGGINGFACEHUB_API_TOKEN=$HF_TOKEN -e LOGFLAG=True opea/llm-sum-tgi:comps
|
||||
|
||||
# check whether tgi is fully ready
|
||||
n=0
|
||||
until [[ "$n" -ge 100 ]] || [[ $ready == true ]]; do
|
||||
docker logs test-comps-llm-sum-tgi-endpoint > ${LOG_PATH}/test-comps-llm-sum-tgi-endpoint.log
|
||||
n=$((n+1))
|
||||
if grep -q Connected ${LOG_PATH}/test-comps-llm-sum-tgi-endpoint.log; then
|
||||
break
|
||||
fi
|
||||
sleep 5s
|
||||
done
|
||||
sleep 5s
|
||||
}
|
||||
|
||||
function validate_services() {
|
||||
local URL="$1"
|
||||
local EXPECTED_RESULT="$2"
|
||||
local SERVICE_NAME="$3"
|
||||
local DOCKER_NAME="$4"
|
||||
local INPUT_DATA="$5"
|
||||
|
||||
local HTTP_STATUS=$(curl -s -o /dev/null -w "%{http_code}" -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL")
|
||||
|
||||
echo "==========================================="
|
||||
|
||||
if [ "$HTTP_STATUS" -eq 200 ]; then
|
||||
echo "[ $SERVICE_NAME ] HTTP status is 200. Checking content..."
|
||||
|
||||
local CONTENT=$(curl -s -X POST -d "$INPUT_DATA" -H 'Content-Type: application/json' "$URL" | tee ${LOG_PATH}/${SERVICE_NAME}.log)
|
||||
|
||||
echo $CONTENT
|
||||
|
||||
if echo "$CONTENT" | grep -q "$EXPECTED_RESULT"; then
|
||||
echo "[ $SERVICE_NAME ] Content is as expected."
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] Content does not match the expected result"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
else
|
||||
echo "[ $SERVICE_NAME ] HTTP status is not 200. Received status was $HTTP_STATUS"
|
||||
docker logs ${DOCKER_NAME} >> ${LOG_PATH}/${SERVICE_NAME}.log
|
||||
exit 1
|
||||
fi
|
||||
sleep 1s
|
||||
}
|
||||
|
||||
function validate_microservices() {
|
||||
sum_port=5076
|
||||
URL="http://${ip_address}:$sum_port/v1/chat/docsum"
|
||||
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"test-comps-llm-sum-tgi-server" \
|
||||
'{"query": "What is Deep Learning?"}'
|
||||
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"test-comps-llm-sum-tgi-server" \
|
||||
'{"query": "What is Deep Learning?", "summary_type": "truncate"}'
|
||||
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"test-comps-llm-sum-tgi-server" \
|
||||
'{"query": "What is Deep Learning?", "summary_type": "map_reduce"}'
|
||||
|
||||
validate_services \
|
||||
"$URL" \
|
||||
'text' \
|
||||
"llm_summarization" \
|
||||
"test-comps-llm-sum-tgi-server" \
|
||||
'{"query": "What is Deep Learning?", "summary_type": "refine"}'
|
||||
}
|
||||
|
||||
function stop_docker() {
|
||||
cid=$(docker ps -aq --filter "name=test-comps-llm-sum-tgi*")
|
||||
if [[ ! -z "$cid" ]]; then docker stop $cid && docker rm $cid && sleep 1s; fi
|
||||
}
|
||||
|
||||
function main() {
|
||||
|
||||
stop_docker
|
||||
|
||||
build_docker_images
|
||||
start_service
|
||||
|
||||
validate_microservices
|
||||
|
||||
stop_docker
|
||||
echo y | docker system prune
|
||||
|
||||
}
|
||||
|
||||
main
|
||||
Reference in New Issue
Block a user