From e236b665dce459d9d11293fbfca03dff66ecba52 Mon Sep 17 00:00:00 2001 From: yuehuazhang Date: Thu, 25 Sep 2025 19:58:14 +0800 Subject: [PATCH] feature:add chart and readme --- incubator/llama-factory-on-tke/Chart.yaml | 36 +++++---- incubator/llama-factory-on-tke/README.md | 93 +++++++++++++++++++++++ 2 files changed, 115 insertions(+), 14 deletions(-) create mode 100644 incubator/llama-factory-on-tke/README.md diff --git a/incubator/llama-factory-on-tke/Chart.yaml b/incubator/llama-factory-on-tke/Chart.yaml index 6be091e..fc2eae7 100644 --- a/incubator/llama-factory-on-tke/Chart.yaml +++ b/incubator/llama-factory-on-tke/Chart.yaml @@ -1,7 +1,26 @@ apiVersion: v2 -name: llama-factory-on-tke +# This is the version number of the application being deployed. This version number should be +# incremented each time you make changes to the application. Versions are not expected to +# follow Semantic Versioning. They should reflect the version the application is using. +# It is recommended to use it with quotes. +appVersion: "1.16.0" description: A Helm chart for Kubernetes - +home: https://github.com/hiyouga/LLaMA-Factory +icon: https://cloudcache.tencent-cloud.com/qcloud/ui/static/Industry_tke/48bef83f-57d3-460c-bdbe-e8e1c57318e1.png +keywords: + - category:AI + - LLaMA-Factory + - tke +maintainers: + - email: yuehuazhang@tencent.com + name: yuehuazhang +name: llama-factory-on-tke +sources: + - https://github.com/bitnami/bitnami-docker-redis +# This is the chart version. This version number should be incremented each time you make changes +# to the chart and its templates, including the app version. +# Versions are expected to follow Semantic Versioning (https://semver.org/) +version: 0.1.0 # A chart can be either an 'application' or a 'library' chart. # # Application charts are a collection of templates that can be packaged into versioned archives @@ -10,15 +29,4 @@ description: A Helm chart for Kubernetes # Library charts provide useful utilities or functions for the chart developer. They're included as # a dependency of application charts to inject those utilities and functions into the rendering # pipeline. Library charts do not define any templates and therefore cannot be deployed. -type: application - -# This is the chart version. This version number should be incremented each time you make changes -# to the chart and its templates, including the app version. -# Versions are expected to follow Semantic Versioning (https://semver.org/) -version: 0.1.0 - -# This is the version number of the application being deployed. This version number should be -# incremented each time you make changes to the application. Versions are not expected to -# follow Semantic Versioning. They should reflect the version the application is using. -# It is recommended to use it with quotes. -appVersion: "1.16.0" +type: application \ No newline at end of file diff --git a/incubator/llama-factory-on-tke/README.md b/incubator/llama-factory-on-tke/README.md new file mode 100644 index 0000000..cc0b806 --- /dev/null +++ b/incubator/llama-factory-on-tke/README.md @@ -0,0 +1,93 @@ +# llama-factory-on-tke + +`llama-factory-on-tke` 是用于在 TKE 上快速部署 [LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory) 的 helm chart。 + +## 架构说明 + +[LLaMA-Factory](https://github.com/hiyouga/LLaMA-Factory) 是一个统一的大语言模型微调框架，支持多种模型的高效微调、推理和评估。在 TKE 部署中，`llama-factory-on-tke` 提供了以下功能： + +- **Web UI**：提供友好的 Web 界面，用于单节点模型微调、推理和管理 +- **API 服务**：提供 RESTful API 接口，支持程序化调用 +- **GPU 支持**：原生支持 NVIDIA GPU 加速单节点训练和推理 +- **模型管理**：支持多种预训练模型的加载和微调 + +该 Chart 将 LLaMA-Factory 部署为单个 Deployment，专注于单节点训练场景，同时暴露 Web UI 和 API 两个服务端口。 + +## 配置说明 + +以下是主要可配置的参数及默认值： + +### 镜像配置 +| Key | Description | Default | +|-------------------|-----------------------------------------------------------------|-----------------------------------------------------| +| image.repository | Docker image repository for the LLaMA-Factory component. | ccr.ccs.tencentyun.com/tke-market/llamafactory | +| image.tag | Tag version of the LLaMA-Factory Docker image. | v0.9.3-0924 | +| image.pullPolicy | Image pull policy for LLaMA-Factory containers. | IfNotPresent | +| imagePullSecrets | Kubernetes secrets for authenticating private image registries. | [] | + +### Service 配置 +| Key | Description | Default | +|--------------------|------------------------------------------------------|---------------| +| service.type | Network service type for the LLaMA-Factory service. | LoadBalancer | +| service.portWebUi | Exposed port number for the Web UI service. | 7860 | +| service.portApi | Exposed port number for the API service. | 8000 | + +### 资源配置 +| Key | Description | Default | +|----------------------------|------------------------------------------------------|----------| +| resources.limits.cpu | CPU resource limit. | 4 | +| resources.limits.memory | Memory resource limit. | 20Gi | +| resources.limits.nvidia.com/gpu | GPU resource limit. | "1" | +| resources.requests.cpu | CPU resource request. | 2 | +| resources.requests.memory | Memory resource request. | 10Gi | +| resources.requests.nvidia.com/gpu | GPU resource request. | "1" | + +### 环境变量配置 +| Key | Description | Default | +|--------------------|------------------------------------------------------|--------------------------| +| env[0].name | Python package index URL environment variable name. | PIP_INDEX | +| env[0].value | Python package index URL. | https://pypi.org/simple | +| env[1].name | Extra features to install environment variable name.| EXTRAS | +| env[1].value | Extra features to install (e.g., metrics). | metrics | + +### 启动命令配置 +| Key | Description | Default | +|---------|------------------------------------------------|------------------------| +| command | Container startup command. | ["bash", "-c"] | +| args | Container startup arguments. | ["llamafactory-cli webui"] | + +### Ingress 配置 +| Key | Description | Default | +|------------------|------------------------------------------------------|---------| +| ingress.enabled | Controls whether to enable Ingress. | false | +| ingress.className| Ingress class name. | "" | +| ingress.hosts | Ingress host configuration. | See `values.yaml` | + +**TKE Ingress 使用注意事项**： +- 如果集群网络模式未采用负载均衡直连 Pod 模式，后端服务的访问类型不能为 ClusterIP +- 建议使用 NodePort 或 LoadBalancer 类型的 Service +- 负载均衡直连 Pod 模式详见：[TKE 文档](https://cloud.tencent.com/document/product/457/41897) + +**Ingress 配置示例**： +```yaml +ingress: + enabled: true + className: "" + hosts: + - host: llama-factory.example.com + paths: + - path: / + pathType: ImplementationSpecific +service: + type: NodePort # 确保与 Ingress 兼容 +``` + +## 示例 + +### 部署 + +使用自定义配置部署 LLaMA-Factory： + +```bash +helm install llama-factory . -f values.yaml +``` \ No newline at end of file