TGI LLM Microservice¶

Text Generation Inference (TGI) is a toolkit for deploying and serving Large Language Models (LLMs). TGI enables high-performance text generation for the most popular open-source LLMs, including Llama, Falcon, StarCoder, BLOOM, GPT-NeoX, and more.

🚀1. Start Microservice with Python (Option 1)¶

To start the LLM microservice, you need to install python packages first.

1.1 Install Requirements¶

pip install -r requirements.txt

1.2 Start LLM Service¶

export HF_TOKEN=${your_hf_api_token}
docker run -p 8008:80 -v ./data:/data --name tgi_service --shm-size 1g ghcr.io/huggingface/text-generation-inference:2.1.0 --model-id ${your_hf_llm_model}

1.3 Verify the TGI Service¶

curl http://${your_ip}:8008/generate \
  -X POST \
  -d '{"inputs":"What is Deep Learning?","parameters":{"max_new_tokens":17, "do_sample": true}}' \
  -H 'Content-Type: application/json'

1.4 Start LLM Service with Python Script¶

export TGI_LLM_ENDPOINT="http://${your_ip}:8008"
python llm.py

🚀2. Start Microservice with Docker (Option 2)¶

If you start an LLM microservice with docker, the docker_compose_llm.yaml file will automatically start a TGI/vLLM service with docker.

2.1 Setup Environment Variables¶

In order to start TGI and LLM services, you need to setup the following environment variables first.

export HF_TOKEN=${your_hf_api_token}
export TGI_LLM_ENDPOINT="http://${your_ip}:8008"
export LLM_MODEL_ID=${your_hf_llm_model}

2.2 Build Docker Image¶

cd ../../../../
docker build -t opea/llm-tgi:latest --build-arg https_proxy=$https_proxy --build-arg http_proxy=$http_proxy -f comps/llms/text-generation/tgi/Dockerfile .

To start a docker container, you have two options:

A. Run Docker with CLI
B. Run Docker with Docker Compose

You can choose one as needed.

2.3 Run Docker with CLI (Option A)¶

docker run -d --name="llm-tgi-server" -p 9000:9000 --ipc=host -e http_proxy=$http_proxy -e https_proxy=$https_proxy -e TGI_LLM_ENDPOINT=$TGI_LLM_ENDPOINT -e HF_TOKEN=$HF_TOKEN opea/llm-tgi:latest

2.4 Run Docker with Docker Compose (Option B)¶

cd text-generation/tgi
docker compose -f docker_compose_llm.yaml up -d

🚀3. Consume LLM Service¶

3.1 Check Service Status¶

curl http://${your_ip}:9000/v1/health_check\
  -X GET \
  -H 'Content-Type: application/json'

3.2 Consume LLM Service¶

You can set the following model parameters according to your actual needs, such as max_tokens, streaming.

The streaming parameter determines the format of the data returned by the API. It will return text string with streaming=false, return text streaming flow with streaming=true.

# non-streaming mode
curl http://${your_ip}:9000/v1/chat/completions \
  -X POST \
  -d '{"query":"What is Deep Learning?","max_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":false}' \
  -H 'Content-Type: application/json'

# streaming mode
curl http://${your_ip}:9000/v1/chat/completions \
  -X POST \
  -d '{"query":"What is Deep Learning?","max_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"repetition_penalty":1.03,"streaming":true}' \
  -H 'Content-Type: application/json'

# consume with SearchedDoc
curl http://${your_ip}:9000/v1/chat/completions \
  -X POST \
  -d '{"initial_query":"What is Deep Learning?","retrieved_docs":[{"text":"Deep Learning is a ..."},{"text":"Deep Learning is b ..."}]}' \
  -H 'Content-Type: application/json'

For parameters in above modes, please refer to HuggingFace InferenceClient API (except we rename ‘max_new_tokens’ to ‘max_tokens’)

# custom chat template
curl http://${your_ip}:9000/v1/chat/completions \
  -X POST \
  -d '{"query":"What is Deep Learning?","max_tokens":17,"top_k":10,"top_p":0.95,"typical_p":0.95,"temperature":0.01,"presence_penalty":1.03", frequency_penalty":0.0, "streaming":true, "chat_template":"### You are a helpful, respectful and honest assistant to help the user with questions.\n### Context: {context}\n### Question: {question}\n### Answer:"}' \
  -H 'Content-Type: application/json'

For parameters in Chat mode, please refer to OpenAI API

4. Validated Model¶

Model	TGI
Intel/neural-chat-7b-v3-3	✓
Llama-2-7b-chat-hf	✓
Llama-2-70b-chat-hf	✓
Meta-Llama-3-8B-Instruct	✓
Meta-Llama-3-70B-Instruct	✓
Phi-3	✓