Skip to content

Latest commit

 

History

History
 
 

CodeGen

README.md

Code Generation Application

Code Generation (CodeGen) Large Language Models (LLMs) are specialized AI models designed for the task of generating computer code. Such models undergo training with datasets that encompass repositories, specialized documentation, programming code, relevant web content, and other related data. They possess a deep understanding of various programming languages, coding patterns, and software development concepts. CodeGen LLMs are engineered to assist developers and programmers. When these LLMs are seamlessly integrated into the developer's Integrated Development Environment (IDE), they possess a comprehensive understanding of the coding context, which includes elements such as comments, function names, and variable names. This contextual awareness empowers them to provide more refined and contextually relevant coding suggestions. Additionally Retrieval-Augmented Generation (RAG) and Agents are parts of the CodeGen example which provide an additional layer of intelligence and adaptability, ensuring that the generated code is not only relevant but also accurate, efficient, and tailored to the specific needs of the developers and programmers.

The capabilities of CodeGen LLMs include:

  • Code Generation: Streamline coding through Code Generation, enabling non-programmers to describe tasks for code creation.
  • Code Completion: Accelerate coding by suggesting contextually relevant snippets as developers type.
  • Code Translation and Modernization: Translate and modernize code across multiple programming languages, aiding interoperability and updating legacy projects.
  • Code Summarization: Extract key insights from codebases, improving readability and developer productivity.
  • Code Refactoring: Offer suggestions for code refactoring, enhancing code performance and efficiency.
  • AI-Assisted Testing: Assist in creating test cases, ensuring code robustness and accelerating development cycles.
  • Error Detection and Debugging: Detect errors in code and provide detailed descriptions and potential fixes, expediting debugging processes.

In this example, we present a Code Copilot application to showcase how code generation can be executed on either Intel Gaudi2 platform or Intel Xeon Processor platform. This CodeGen use case involves code generation utilizing open-source models such as m-a-p/OpenCodeInterpreter-DS-6.7B and deepseek-ai/deepseek-coder-33b-instruct with Text Generation Inference (TGI) for serving deployment.

The workflow falls into the following architecture:

architecture

The CodeGen example is implemented using the component-level microservices defined in GenAIComps. The flow chart below shows the information flow between different microservices for this example.

---
config:
  flowchart:
    nodeSpacing: 400
    rankSpacing: 100
    curve: linear
  themeVariables:
    fontSize: 25px
---
flowchart LR
    %% Colors %%
    classDef blue fill:#ADD8E6,stroke:#ADD8E6,stroke-width:2px,fill-opacity:0.5
    classDef orange fill:#FBAA60,stroke:#ADD8E6,stroke-width:2px,fill-opacity:0.5
    classDef orchid fill:#C26DBC,stroke:#ADD8E6,stroke-width:2px,fill-opacity:0.5
    classDef invisible fill:transparent,stroke:transparent;
    style CodeGen-MegaService stroke:#000000
    %% Subgraphs %%
    subgraph CodeGen-MegaService["CodeGen-MegaService"]
        direction LR
        EM([Embedding<br>MicroService]):::blue
        RET([Retrieval<br>MicroService]):::blue
        RER([Agents]):::blue
        LLM([LLM<br>MicroService]):::blue
    end
    subgraph User Interface
        direction LR
        a([Submit Query Tab]):::orchid
        UI([UI server]):::orchid
        Ingest([Manage Resources]):::orchid
    end

    CLIP_EM{{Embedding<br>service}}
    VDB{{Vector DB}}
    V_RET{{Retriever<br>service}}
    Ingest{{Ingest data}}
    DP([Data Preparation]):::blue
    LLM_gen{{TGI Service}}
    GW([CodeGen GateWay]):::orange

    %% Data Preparation flow
    %% Ingest data flow
    direction LR
    Ingest[Ingest data] --> UI
    UI --> DP
    DP <-.-> CLIP_EM

    %% Questions interaction
    direction LR
    a[User Input Query] --> UI
    UI --> GW
    GW <==> CodeGen-MegaService
    EM ==> RET
    RET ==> RER
    RER ==> LLM


    %% Embedding service flow
    direction LR
    EM <-.-> CLIP_EM
    RET <-.-> V_RET
    LLM <-.-> LLM_gen

    direction TB
    %% Vector DB interaction
    V_RET <-.->VDB
    DP <-.->VDB
Loading

🤖 Automated Terraform Deployment using Intel® Optimized Cloud Modules for Terraform

Cloud Provider Intel Architecture Intel Optimized Cloud Module for Terraform Comments
AWS 4th Gen Intel Xeon with Intel AMX AWS Deployment
GCP 4th/5th Gen Intel Xeon GCP Deployment
Azure 4th/5th Gen Intel Xeon Work-in-progress
Intel Tiber AI Cloud 5th Gen Intel Xeon with Intel AMX Work-in-progress

Manual Deployment of CodeGen Service

The CodeGen service can be effortlessly deployed on either Intel Gaudi2 or Intel Xeon Scalable Processor.

Currently we support two ways of deploying CodeGen services with docker compose:

  1. Start services using the docker image on docker hub:

    docker pull opea/codegen:latest

  2. Start services using the docker images built from source. See the Gaudi Guide or Xeon Guide for more information.

Required Models

By default, the LLM model is set to a default value as listed below:

Service Model
LLM_MODEL_ID Qwen/Qwen2.5-Coder-32B-Instruct

Qwen/Qwen2.5-Coder-32B-Instruct may be a gated model that requires submitting an access request through Hugging Face. You can replace it with another model for m. Change the LLM_MODEL_ID below for your needs, such as: Qwen/Qwen2.5-Coder-7B-Instruct, deepseek-ai/deepseek-coder-6.7b-instruct

If you choose to use meta-llama/CodeLlama-7b-hf as LLM model, you will need to visit here, click the Expand to review and access button to ask for model access.

Setup Environment Variable

To set up environment variables for deploying CodeGen services, follow these steps:

  1. Set the required environment variables:

    # Example: host_ip="192.168.1.1"
export host_ip="External_Public_IP"
# Example: no_proxy="localhost, 127.0.0.1, 192.168.1.1"
export no_proxy="Your_No_Proxy"
export HUGGINGFACEHUB_API_TOKEN="Your_Huggingface_API_Token"

  2. If you are in a proxy environment, also set the proxy-related environment variables:

    export http_proxy="Your_HTTP_Proxy"
export https_proxy="Your_HTTPs_Proxy"

  3. Set up other environment variables:

    source ./docker_compose/set_env.sh

Deploy CodeGen using Docker

Deploy CodeGen on Gaudi

Find the corresponding compose.yaml. User could start CodeGen based on TGI or vLLM service:

cd GenAIExamples/CodeGen/docker_compose/intel/hpu/gaudi

TGI service:

docker compose --profile codegen-gaudi-tgi up -d

vLLM service:

docker compose --profile codegen-gaudi-vllm up -d

Refer to the Gaudi Guide to build docker images from source.

Deploy CodeGen on Xeon

Find the corresponding compose.yaml. User could start CodeGen based on TGI or vLLM service:

cd GenAIExamples/CodeGen/docker_compose/intel/cpu/xeon

TGI service:

docker compose --profile codegen-xeon-tgi up -d

vLLM service:

docker compose --profile codegen-xeon-vllm up -d

Refer to the Xeon Guide for more instructions on building docker images from source.

Deploy CodeGen on Kubernetes using Helm Chart

Refer to the CodeGen helm chart for instructions on deploying CodeGen on Kubernetes.

Consume CodeGen Service

Two ways of consuming CodeGen Service:

  1. Use cURL command on terminal

    curl http://${host_ip}:7778/v1/codegen \
    -H "Content-Type: application/json" \
    -d '{"messages": "Implement a high-level API for a TODO list application. The API takes as input an operation request and updates the TODO list in place. If the request is invalid, raise an exception."}'

    If the user wants a CodeGen service with RAG and Agents based on dedicated documentation.

    curl http://localhost:7778/v1/codegen \
   -H "Content-Type: application/json" \
   -d '{"agents_flag": "True", "index_name": "my_API_document", "messages": "Implement a high-level API for a TODO list application. The API takes as input an operation request and updates the TODO list in place. If the request is invalid, raise an exception."}'

  2. Access via frontend

    To access the frontend, open the following URL in your browser: http://{host_ip}:5173.

    By default, the UI runs on port 5173 internally.

Troubleshooting

  1. If you get errors like "Access Denied", validate micro service first. A simple example:

    http_proxy=""
curl http://${host_ip}:8028/generate \
  -X POST \
  -d '{"inputs":"Implement a high-level API for a TODO list application. The API takes as input an operation request and updates the TODO list in place. If the request is invalid, raise an exception.","parameters":{"max_tokens":256, "do_sample": true}}' \
  -H 'Content-Type: application/json'

  2. If you get errors like "aiohttp.client_exceptions.ClientConnectorError: Cannot connect to host xx.xx.xx.xx:8028", check the tgi service first. If there is "Cannot access gated repo for url https://huggingface.co/meta-llama/CodeLlama-7b-hf/resolve/main/config.json." error of tgi service, Then you need to ask for model access first. Follow the instruction in the Required Models section for more information.

  3. (Docker only) If all microservices work well, check the port ${host_ip}:7778, the port may be allocated by other users, you can modify the compose.yaml.

  4. (Docker only) If you get errors like "The container name is in use", change container name in compose.yaml.