Update AgentQnA and DocIndexRetriever

AgentQnA/README.md

@@ -84,7 +84,7 @@ flowchart LR
 3. Hierarchical multi-agents can improve performance.
    Expert worker agents, such as RAG agent and SQL agent, can provide high-quality output for different aspects of a complex query, and the supervisor agent can aggregate the information together to provide a comprehensive answer. If we only use one agent and provide all the tools to this single agent, it may get overwhelmed and not able to provide accurate answers.
 
-## Deployment with docker
+## Deploy with docker
 
 1. Build agent docker image [Optional]
 
@@ -217,13 +217,19 @@ docker build -t opea/agent:latest --build-arg https_proxy=$https_proxy --build-a
    :::
    ::::
 
+## Deploy AgentQnA UI
+
+The AgentQnA UI can be deployed locally or using Docker.
+
+For detailed instructions on deploying AgentQnA UI, refer to the [AgentQnA UI Guide](./ui/svelte/README.md).
+
 ## Deploy using Helm Chart
 
 Refer to the [AgentQnA helm chart](./kubernetes/helm/README.md) for instructions on deploying AgentQnA on Kubernetes.
 
 ## Validate services
 
-First look at logs of the agent docker containers:
+1. First look at logs of the agent docker containers:
 
 ```
 # worker RAG agent
@@ -240,35 +246,18 @@ docker logs react-agent-endpoint
 
 You should see something like "HTTP server setup successful" if the docker containers are started successfully.</p>
 
-Second, validate worker RAG agent:
+2. You can use python to validate the agent system
 
-```
-curl http://${host_ip}:9095/v1/chat/completions -X POST -H "Content-Type: application/json" -d '{
-     "messages": "Michael Jackson song Thriller"
-    }'
-```
+```bash
+# RAG worker agent
+python tests/test.py --prompt "Tell me about Michael Jackson song Thriller" --agent_role "worker" --ext_port 9095
 
-Third, validate worker SQL agent:
+# SQL agent
+python tests/test.py --prompt "How many employees in company" --agent_role "worker" --ext_port 9096
 
+# supervisor agent: this will test a two-turn conversation
+python tests/test.py --agent_role "supervisor" --ext_port 9090
 ```
-curl http://${host_ip}:9096/v1/chat/completions -X POST -H "Content-Type: application/json" -d '{
-     "messages": "How many employees are in the company"
-    }'
-```
-
-Finally, validate supervisor agent:
-
-```
-curl http://${host_ip}:9090/v1/chat/completions -X POST -H "Content-Type: application/json" -d '{
-     "messages": "How many albums does Iron Maiden have?"
-    }'
-```
-
-## Deploy AgentQnA UI
-
-The AgentQnA UI can be deployed locally or using Docker.
-
-For detailed instructions on deploying AgentQnA UI, refer to the [AgentQnA UI Guide](./ui/svelte/README.md).
 
 ## How to register your own tools with agent
 

AgentQnA/docker_compose/intel/cpu/xeon/compose_openai.yaml

@@ -13,6 +13,7 @@ services:
     environment:
       ip_address: ${ip_address}
       strategy: rag_agent
+      with_memory: false
       recursion_limit: ${recursion_limit_worker}
       llm_engine: openai
       OPENAI_API_KEY: ${OPENAI_API_KEY}
@@ -35,17 +36,17 @@ services:
     image: opea/agent:latest
     container_name: sql-agent-endpoint
     volumes:
-      - ${WORKDIR}/TAG-Bench/:/home/user/TAG-Bench # SQL database
+      - ${WORKDIR}/GenAIExamples/AgentQnA/tests:/home/user/chinook-db # SQL database
     ports:
       - "9096:9096"
     ipc: host
     environment:
       ip_address: ${ip_address}
       strategy: sql_agent
+      with_memory: false
       db_name: ${db_name}
       db_path: ${db_path}
       use_hints: false
-      hints_file: /home/user/TAG-Bench/${db_name}_hints.csv
       recursion_limit: ${recursion_limit_worker}
       llm_engine: openai
       OPENAI_API_KEY: ${OPENAI_API_KEY}
@@ -64,21 +65,23 @@ services:
     container_name: react-agent-endpoint
     depends_on:
       - worker-rag-agent
+      - worker-sql-agent
     volumes:
       - ${TOOLSET_PATH}:/home/user/tools/
     ports:
       - "9090:9090"
     ipc: host
     environment:
       ip_address: ${ip_address}
-      strategy: react_langgraph
+      strategy: react_llama
+      with_memory: true
       recursion_limit: ${recursion_limit_supervisor}
       llm_engine: openai
       OPENAI_API_KEY: ${OPENAI_API_KEY}
       model: ${model}
       temperature: ${temperature}
       max_new_tokens: ${max_new_tokens}
-      stream: false
+      stream: true
       tools: /home/user/tools/supervisor_agent_tools.yaml
       require_human_feedback: false
       no_proxy: ${no_proxy}

AgentQnA/docker_compose/intel/cpu/xeon/launch_agent_service_openai.sh

@@ -16,7 +16,7 @@ export WORKER_AGENT_URL="http://${ip_address}:9095/v1/chat/completions"
 export SQL_AGENT_URL="http://${ip_address}:9096/v1/chat/completions"
 export RETRIEVAL_TOOL_URL="http://${ip_address}:8889/v1/retrievaltool"
 export CRAG_SERVER=http://${ip_address}:8080
-export db_name=california_schools
-export db_path="sqlite:////home/user/TAG-Bench/dev_folder/dev_databases/${db_name}/${db_name}.sqlite"
+export db_name=Chinook
+export db_path="sqlite:////home/user/chinook-db/Chinook_Sqlite.sqlite"
 
 docker compose -f compose_openai.yaml up -d

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

@@ -13,6 +13,7 @@ services:
     environment:
       ip_address: ${ip_address}
       strategy: rag_agent_llama
+      with_memory: false
       recursion_limit: ${recursion_limit_worker}
       llm_engine: vllm
       HUGGINGFACEHUB_API_TOKEN: ${HUGGINGFACEHUB_API_TOKEN}
@@ -43,6 +44,7 @@ services:
     environment:
       ip_address: ${ip_address}
       strategy: sql_agent_llama
+      with_memory: false
       db_name: ${db_name}
       db_path: ${db_path}
       use_hints: false
@@ -74,14 +76,15 @@ services:
     environment:
       ip_address: ${ip_address}
       strategy: react_llama
+      with_memory: true
       recursion_limit: ${recursion_limit_supervisor}
       llm_engine: vllm
       HUGGINGFACEHUB_API_TOKEN: ${HUGGINGFACEHUB_API_TOKEN}
       llm_endpoint_url: ${LLM_ENDPOINT_URL}
       model: ${LLM_MODEL_ID}
       temperature: ${temperature}
       max_new_tokens: ${max_new_tokens}
-      stream: false
+      stream: true
       tools: /home/user/tools/supervisor_agent_tools.yaml
       require_human_feedback: false
       no_proxy: ${no_proxy}

AgentQnA/docker_compose/intel/hpu/gaudi/launch_agent_service_gaudi.sh

@@ -14,7 +14,7 @@ export HUGGINGFACEHUB_API_TOKEN=${HUGGINGFACEHUB_API_TOKEN}
 export HF_CACHE_DIR=${HF_CACHE_DIR}
 ls $HF_CACHE_DIR
 export HUGGINGFACEHUB_API_TOKEN=${HUGGINGFACEHUB_API_TOKEN}
-export LLM_MODEL_ID="meta-llama/Meta-Llama-3.1-70B-Instruct"
+export LLM_MODEL_ID="meta-llama/Llama-3.3-70B-Instruct" #"meta-llama/Meta-Llama-3.1-70B-Instruct"
 export NUM_SHARDS=4
 export LLM_ENDPOINT_URL="http://${ip_address}:8086"
 export temperature=0

AgentQnA/tests/step4_launch_and_validate_agent_tgi.sh → AgentQnA/tests/step4_launch_and_validate_agent_gaudi.sh

@@ -11,7 +11,7 @@ export ip_address=$(hostname -I | awk '{print $1}')
 export TOOLSET_PATH=$WORKPATH/tools/
 export HUGGINGFACEHUB_API_TOKEN=${HUGGINGFACEHUB_API_TOKEN}
 HF_TOKEN=${HUGGINGFACEHUB_API_TOKEN}
-model="meta-llama/Meta-Llama-3.1-70B-Instruct"
+model="meta-llama/Llama-3.3-70B-Instruct" #"meta-llama/Meta-Llama-3.1-70B-Instruct"
 
 export HF_CACHE_DIR=/data2/huggingface
 if [ ! -d "$HF_CACHE_DIR" ]; then
@@ -60,23 +60,6 @@ function start_vllm_service_70B() {
     echo "Service started successfully"
 }
 
-
-function prepare_data() {
-    cd $WORKDIR
-
-    echo "Downloading data..."
-    git clone https://github.com/TAG-Research/TAG-Bench.git
-    cd TAG-Bench/setup
-    chmod +x get_dbs.sh
-    ./get_dbs.sh
-
-    echo "Split data..."
-    cd $WORKPATH/tests/sql_agent_test
-    bash run_data_split.sh
-
-    echo "Data preparation done!"
-}
-
 function download_chinook_data(){
     echo "Downloading chinook data..."
     cd $WORKDIR
@@ -113,7 +96,7 @@ function validate_agent_service() {
     echo "======================Testing worker rag agent======================"
     export agent_port="9095"
     prompt="Tell me about Michael Jackson song Thriller"
-    local CONTENT=$(python3 $WORKDIR/GenAIExamples/AgentQnA/tests/test.py --prompt "$prompt")
+    local CONTENT=$(python3 $WORKDIR/GenAIExamples/AgentQnA/tests/test.py --prompt "$prompt" --agent_role "worker" --ext_port $agent_port)
     # echo $CONTENT
     local EXIT_CODE=$(validate "$CONTENT" "Thriller" "rag-agent-endpoint")
     echo $EXIT_CODE
@@ -127,7 +110,7 @@ function validate_agent_service() {
     echo "======================Testing worker sql agent======================"
     export agent_port="9096"
     prompt="How many employees are there in the company?"
-    local CONTENT=$(python3 $WORKDIR/GenAIExamples/AgentQnA/tests/test.py --prompt "$prompt")
+    local CONTENT=$(python3 $WORKDIR/GenAIExamples/AgentQnA/tests/test.py --prompt "$prompt" --agent_role "worker" --ext_port $agent_port)
     local EXIT_CODE=$(validate "$CONTENT" "8" "sql-agent-endpoint")
     echo $CONTENT
     # echo $EXIT_CODE
@@ -140,9 +123,8 @@ function validate_agent_service() {
     # test supervisor react agent
     echo "======================Testing supervisor react agent======================"
     export agent_port="9090"
-    prompt="How many albums does Iron Maiden have?"
-    local CONTENT=$(python3 $WORKDIR/GenAIExamples/AgentQnA/tests/test.py --prompt "$prompt")
-    local EXIT_CODE=$(validate "$CONTENT" "21" "react-agent-endpoint")
+    local CONTENT=$(python3 $WORKDIR/GenAIExamples/AgentQnA/tests/test.py --agent_role "supervisor" --ext_port $agent_port --stream)
+    local EXIT_CODE=$(validate "$CONTENT" "Iron" "react-agent-endpoint")
     # echo $CONTENT
     echo $EXIT_CODE
     local EXIT_CODE="${EXIT_CODE:0-1}"
@@ -153,15 +135,6 @@ function validate_agent_service() {
 
 }
 
-function remove_data() {
-    echo "Removing data..."
-    cd $WORKDIR
-    if [ -d "TAG-Bench" ]; then
-        rm -rf TAG-Bench
-    fi
-    echo "Data removed!"
-}
-
 function remove_chinook_data(){
     echo "Removing chinook data..."
     cd $WORKDIR
@@ -189,8 +162,9 @@ function main() {
     echo "==================== Agent service validated ===================="
 }
 
-remove_data
+
 remove_chinook_data
+
 main
-remove_data
+
 remove_chinook_data

AgentQnA/tests/test.py

@@ -1,34 +1,20 @@
-# Copyright (C) 2024 Intel Corporation
+# Copyright (C) 2025 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 
 import argparse
-import os
+import json
+import uuid
 
 import requests
 
 
-def generate_answer_agent_api(url, prompt):
-    proxies = {"http": ""}
-    payload = {
-        "messages": prompt,
-    }
-    response = requests.post(url, json=payload, proxies=proxies)
-    answer = response.json()["text"]
-    return answer
-
-
 def process_request(url, query, is_stream=False):
     proxies = {"http": ""}
-
-    payload = {
-        "messages": query,
-    }
-
+    content = json.dumps(query) if query is not None else None
     try:
-        resp = requests.post(url=url, json=payload, proxies=proxies, stream=is_stream)
+        resp = requests.post(url=url, data=content, proxies=proxies, stream=is_stream)
         if not is_stream:
             ret = resp.json()["text"]
-            print(ret)
         else:
             for line in resp.iter_lines(decode_unicode=True):
                 print(line)
@@ -38,19 +24,54 @@ def process_request(url, query, is_stream=False):
         return ret
     except requests.exceptions.RequestException as e:
         ret = f"An error occurred:{e}"
-        print(ret)
-        return False
+        return None
+
+
+def test_worker_agent(args):
+    url = f"http://{args.ip_addr}:{args.ext_port}/v1/chat/completions"
+    query = {"role": "user", "messages": args.prompt, "stream": "false"}
+    ret = process_request(url, query)
+    print("Response: ", ret)
+
+
+def add_message_and_run(url, user_message, thread_id, stream=False):
+    print("User message: ", user_message)
+    query = {"role": "user", "messages": user_message, "thread_id": thread_id, "stream": stream}
+    ret = process_request(url, query, is_stream=stream)
+    print("Response: ", ret)
+
+
+def test_chat_completion_multi_turn(args):
+    url = f"http://{args.ip_addr}:{args.ext_port}/v1/chat/completions"
+    thread_id = f"{uuid.uuid4()}"
+
+    # first turn
+    print("===============First turn==================")
+    user_message = "Which artist has the most albums in the database?"
+    add_message_and_run(url, user_message, thread_id, stream=args.stream)
+    print("===============End of first turn==================")
+
+    # second turn
+    print("===============Second turn==================")
+    user_message = "Give me a few examples of the artist's albums?"
+    add_message_and_run(url, user_message, thread_id, stream=args.stream)
+    print("===============End of second turn==================")
 
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("--prompt", type=str)
-    parser.add_argument("--stream", action="store_true")
-    args = parser.parse_args()
+    parser.add_argument("--ip_addr", type=str, default="127.0.0.1", help="endpoint ip address")
+    parser.add_argument("--ext_port", type=str, default="9090", help="endpoint port")
+    parser.add_argument("--stream", action="store_true", help="streaming mode")
+    parser.add_argument("--prompt", type=str, help="prompt message")
+    parser.add_argument("--agent_role", type=str, default="supervisor", help="supervisor or worker")
+    args, _ = parser.parse_known_args()
 
-    ip_address = os.getenv("ip_address", "localhost")
-    agent_port = os.getenv("agent_port", "9090")
-    url = f"http://{ip_address}:{agent_port}/v1/chat/completions"
-    prompt = args.prompt
+    print(args)
 
-    process_request(url, prompt, args.stream)
+    if args.agent_role == "supervisor":
+        test_chat_completion_multi_turn(args)
+    elif args.agent_role == "worker":
+        test_worker_agent(args)
+    else:
+        raise ValueError("Invalid agent role")

AgentQnA/tests/test_compose_on_gaudi.sh

@@ -78,7 +78,7 @@ bash step3_ingest_data_and_validate_retrieval.sh
 echo "=================== #3 Data ingestion and validation completed===================="
 
 echo "=================== #4 Start agent and API server===================="
-bash step4_launch_and_validate_agent_tgi.sh
+bash step4_launch_and_validate_agent_gaudi.sh
 echo "=================== #4 Agent test passed ===================="
 
 echo "=================== #5 Stop agent and API server===================="