状态：待评审。本文档不参与站点构建（位于仓库根目录，
doom只读docs/）。
参照 Red Hat OpenShift AI 3.x 的组织方式，顶层导航按用户场景/生命周期阶段划分，
而非按功能模块或上游组件划分。与 RHOAI 的唯一结构性差异：Alauda AI 的 19 个组件是
独立安装的 operator / cluster plugin，因此每个场景保留一个 components/ 子目录承载
该场景所依赖组件的介绍与安装文档；RHOAI 因组件由 DataScienceCluster 统一开关，无此需要。
组件归属规则：按任务文档的主场景归位，只存放一份。次要场景通过任务页的 Prerequisites 内联链接与章节导语指向，不设置仅含跳转的占位页。
| # | 章节 | 目录 | 页数 | 含组件 |
|---|---|---|---|---|
| 1 | Overview 概述 | overview/ |
6 | — |
| 2 | Plan 规划与选型 | plan/ |
11 | — |
| 3 | Install 安装 | installation/ |
6 | — |
| 4 | Upgrade 升级与卸载 | upgrade/ |
5 | — |
| 5 | Administer 平台管理 | administer/ |
13 | — |
| 6 | Develop 开发 | develop/ |
56 | data_science_pipelines, feast, kubeflow, kuberay, label_studio, mlflow, spark_operator |
| 7 | Train 训练与调优 | train/ |
30 | jobset, kueue, volcano |
| 8 | Deploy 部署与推理 | deploy/ |
48 | envoy_ai_gateway, infernex_bridge, kserve, lws |
| 9 | Build AI Applications 构建 AI 应用 | ai_applications/ |
23 | dify, kagenti, llama_stack, mcp_lifecycle_operator |
| 10 | Evaluate & Safety 评测与安全 | evaluate_safety/ |
9 | trustyai |
| 11 | Monitor 监控与运维 | monitor/ |
11 | — |
| 12 | API Reference | apis/ |
13 | — |
合计 232 页（含 2 篇 .md：plan/supported_configurations.md、train/guides/kubeflow-trainer-quick-start.md，doom 的路由 **/*.md{,x} 一并收录）。
注：本表为第一版（PR #299 首次提交）的映射；其中
how_to//functions/相关条目已被 §8 的二次调整取代。
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
overview/architecture.mdx |
overview/architecture.mdx |
1 | |
overview/index.mdx |
overview/index.mdx |
||
overview/intro.mdx |
overview/intro.mdx |
||
overview/quick_start.mdx |
overview/quick_start.mdx |
1 | |
overview/release_notes.mdx |
overview/release_notes.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
infrastructure_management/device_management/hami.mdx |
plan/device_options/hami.mdx |
||
infrastructure_management/device_management/index.mdx |
plan/device_options/index.mdx |
||
infrastructure_management/device_management/pgpu.mdx |
plan/device_options/pgpu.mdx |
||
learn/glossary.mdx |
plan/glossary.mdx |
||
learn/supported-configurations-for-2.x.md |
plan/supported_configurations.md |
重命名 | |
model_inference/inference_guide/deepseek-v4-flash-w4a8.mdx |
plan/validated_models/deepseek-v4-flash-w4a8.mdx |
||
model_inference/inference_guide/deepseek-v4-flash-w8a8.mdx |
plan/validated_models/deepseek-v4-flash-w8a8.mdx |
||
model_inference/inference_guide/index.mdx |
plan/validated_models/index.mdx |
8 | |
model_inference/inference_guide/minimax-m2.5-w8a8.mdx |
plan/validated_models/minimax-m2.5-w8a8.mdx |
||
model_inference/inference_guide/qwen3-32b.mdx |
plan/validated_models/qwen3-32b.mdx |
||
model_inference/inference_guide/qwen3-6-27b-w8a8.mdx |
plan/validated_models/qwen3-6-27b-w8a8.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
installation/ai-cluster.mdx |
installation/ai-cluster.mdx |
3 | |
installation/ai-essentials.mdx |
installation/ai-essentials.mdx |
||
installation/index.mdx |
installation/index.mdx |
5 | |
installation/pre-configuration.mdx |
installation/pre-configuration.mdx |
||
installation/tools.mdx |
installation/tools.mdx |
||
installation/workbench.mdx |
installation/workbench.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
upgrade/index.mdx |
upgrade/index.mdx |
||
upgrade/migrating-to-knative-operator.mdx |
upgrade/migrating-to-knative-operator.mdx |
1 | |
upgrade/uninstall.mdx |
upgrade/uninstall.mdx |
||
upgrade/upgrade-from-previous-version.mdx |
upgrade/upgrade-from-previous-version.mdx |
6 | |
workbench/upgrade.mdx |
upgrade/workbench.mdx |
移入 Upgrade 章 | 1 |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
infrastructure_management/hardware_profile/functions/hardware_profile.mdx |
administer/hardware_profile/functions/hardware_profile.mdx |
||
infrastructure_management/hardware_profile/functions/index.mdx |
administer/hardware_profile/functions/index.mdx |
||
infrastructure_management/hardware_profile/how_to/cpu_and_gpu_profiles.mdx |
administer/hardware_profile/how_to/cpu_and_gpu_profiles.mdx |
||
infrastructure_management/hardware_profile/how_to/create_hardware_profile_cli.mdx |
administer/hardware_profile/how_to/create_hardware_profile_cli.mdx |
||
infrastructure_management/hardware_profile/how_to/index.mdx |
administer/hardware_profile/how_to/index.mdx |
||
infrastructure_management/hardware_profile/how_to/schedule_to_specific_gpu_nodes.mdx |
administer/hardware_profile/how_to/schedule_to_specific_gpu_nodes.mdx |
||
infrastructure_management/hardware_profile/index.mdx |
administer/hardware_profile/index.mdx |
||
infrastructure_management/hardware_profile/intro.mdx |
administer/hardware_profile/intro.mdx |
||
ai_applications/kagenti/how_to/enable_secure_profile.mdx |
administer/how_to/enable_secure_profile.mdx |
5 | |
ai_applications/kagenti/how_to/install_secure_dependencies.mdx |
administer/how_to/install_secure_dependencies.mdx |
2 | |
data_experiments/mlflow/how_to/workspaces.mdx |
administer/how_to/mlflow_workspaces.mdx |
3 | |
infrastructure_management/multi_tenant/functions/index.mdx |
administer/multi_tenant/functions/index.mdx |
||
infrastructure_management/multi_tenant/functions/namespace-manage.mdx |
administer/multi_tenant/functions/namespace-manage.mdx |
||
infrastructure_management/multi_tenant/index.mdx |
administer/multi_tenant/index.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
training_pipelines/agentic_mlops/coding-agents-with-inference-service.mdx |
develop/agentic_mlops/coding-agents-with-inference-service.mdx |
30 | |
training_pipelines/agentic_mlops/index.mdx |
develop/agentic_mlops/index.mdx |
||
training_pipelines/agentic_mlops/mlops-with-coding-agents.mdx |
develop/agentic_mlops/mlops-with-coding-agents.mdx |
33 | |
training_pipelines/data_science_pipelines/index.mdx |
develop/components/data_science_pipelines/index.mdx |
||
training_pipelines/data_science_pipelines/install.mdx |
develop/components/data_science_pipelines/install.mdx |
2 | |
training_pipelines/data_science_pipelines/intro.mdx |
develop/components/data_science_pipelines/intro.mdx |
★标题 "Introduction" 需改为组件名 | 1 |
data_experiments/feast/index.mdx |
develop/components/feast/index.mdx |
||
data_experiments/feast/install.mdx |
develop/components/feast/install.mdx |
1 | |
data_experiments/feast/intro.mdx |
develop/components/feast/intro.mdx |
★标题 "Introduction" 需改为组件名 | 2 |
data_experiments/feast/quickstart.mdx |
develop/components/feast/quickstart.mdx |
1 | |
training_pipelines/kubeflow/faq.mdx |
develop/components/kubeflow/faq.mdx |
||
training_pipelines/kubeflow/index.mdx |
develop/components/kubeflow/index.mdx |
||
training_pipelines/kubeflow/install.mdx |
develop/components/kubeflow/install.mdx |
1 | |
training_pipelines/kubeflow/intro.mdx |
develop/components/kubeflow/intro.mdx |
★标题 "Introduction" 需改为组件名 | 1 |
training_pipelines/kubeflow/upgrade.mdx |
develop/components/kubeflow/upgrade.mdx |
1 | |
training_pipelines/kuberay/index.mdx |
develop/components/kuberay/index.mdx |
||
training_pipelines/kuberay/install.mdx |
develop/components/kuberay/install.mdx |
||
training_pipelines/kuberay/intro.mdx |
develop/components/kuberay/intro.mdx |
★标题 "Introduction" 需改为组件名 | |
data_experiments/label_studio/overview/features.mdx |
develop/components/label_studio/features.mdx |
打平 overview/ 子目录 | |
data_experiments/label_studio/index.mdx |
develop/components/label_studio/index.mdx |
||
data_experiments/label_studio/install.mdx |
develop/components/label_studio/install.mdx |
||
data_experiments/label_studio/overview/intro.mdx |
develop/components/label_studio/intro.mdx |
打平 overview/ 子目录 ; ★标题 "Introduction" 需改为组件名 | |
data_experiments/label_studio/quickstart.mdx |
develop/components/label_studio/quickstart.mdx |
||
data_experiments/mlflow/index.mdx |
develop/components/mlflow/index.mdx |
||
data_experiments/mlflow/install.mdx |
develop/components/mlflow/install.mdx |
2 | |
data_experiments/mlflow/intro.mdx |
develop/components/mlflow/intro.mdx |
★标题 "Introduction" 需改为组件名 | 3 |
training_pipelines/spark_operator/index.mdx |
develop/components/spark_operator/index.mdx |
||
training_pipelines/spark_operator/install.mdx |
develop/components/spark_operator/install.mdx |
1 | |
training_pipelines/spark_operator/intro.mdx |
develop/components/spark_operator/intro.mdx |
★标题 "Introduction" 需改为组件名 | |
workbench/connections/how_to/index.mdx |
develop/connections/how_to/index.mdx |
||
workbench/connections/how_to/using_connections.mdx |
develop/connections/how_to/using_connections.mdx |
1 | |
workbench/connections/index.mdx |
develop/connections/index.mdx |
||
workbench/connections/overview/index.mdx |
develop/connections/overview/index.mdx |
||
workbench/connections/overview/intro.mdx |
develop/connections/overview/intro.mdx |
1 | |
data_experiments/mlflow/how_to/agent-tracing.mdx |
develop/experiment_tracking/agent-tracing.mdx |
3 | |
data_experiments/mlflow/how_to/mlflow-python-sdk.mdx |
develop/experiment_tracking/mlflow-python-sdk.mdx |
4 | |
data_experiments/mlflow/how_to/pipelines-mlflow-integration.mdx |
develop/experiment_tracking/pipelines-mlflow-integration.mdx |
14 | |
training_pipelines/kuberay/how_to/codeflare-sdk-tutorial.mdx |
develop/how_to/codeflare-sdk-tutorial.mdx |
2 | |
training_pipelines/data_science_pipelines/how_to/create_dspa.mdx |
develop/how_to/create_dspa.mdx |
1 | |
training_pipelines/kubeflow/how_to/model-registry.mdx |
develop/how_to/model-registry.mdx |
||
training_pipelines/kubeflow/how_to/notebooks.mdx |
develop/how_to/notebooks.mdx |
||
training_pipelines/kubeflow/how_to/pipelines.mdx |
develop/how_to/pipelines.mdx |
||
training_pipelines/spark_operator/how_to/run_spark_application.mdx |
develop/how_to/run_spark_application.mdx |
1 | |
training_pipelines/kueue/how_to/tekton.mdx |
develop/how_to/tekton.mdx |
||
training_pipelines/kubeflow/how_to/tensorboards.mdx |
develop/how_to/tensorboards.mdx |
||
model_inference/model_management/how_to/upload_models_using_notebook.mdx |
develop/how_to/upload_models_using_notebook.mdx |
||
training_pipelines/kubeflow/how_to/volumes-kserve.mdx |
develop/how_to/volumes-kserve.mdx |
||
workbench/how_to/create_workbench.mdx |
develop/workbench/how_to/create_workbench.mdx |
||
workbench/how_to/index.mdx |
develop/workbench/how_to/index.mdx |
||
workbench/how_to/run-kubeflow-pipelines-with-elyra.mdx |
develop/workbench/how_to/run-kubeflow-pipelines-with-elyra.mdx |
9 | |
workbench/index.mdx |
develop/workbench/index.mdx |
||
workbench/overview/index.mdx |
develop/workbench/overview/index.mdx |
||
workbench/overview/intro.mdx |
develop/workbench/overview/intro.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
training_pipelines/jobset/index.mdx |
train/components/jobset/index.mdx |
||
training_pipelines/jobset/install.mdx |
train/components/jobset/install.mdx |
1 | |
training_pipelines/jobset/intro.mdx |
train/components/jobset/intro.mdx |
★标题 "Introduction" 需改为组件名 | 3 |
training_pipelines/jobset/quickstart.mdx |
train/components/jobset/quickstart.mdx |
2 | |
training_pipelines/kueue/index.mdx |
train/components/kueue/index.mdx |
||
training_pipelines/kueue/install.mdx |
train/components/kueue/install.mdx |
||
training_pipelines/kueue/intro.mdx |
train/components/kueue/intro.mdx |
★标题 "Introduction" 需改为组件名 | |
training_pipelines/volcano/index.mdx |
train/components/volcano/index.mdx |
||
training_pipelines/volcano/install.mdx |
train/components/volcano/install.mdx |
||
training_pipelines/volcano/intro.mdx |
train/components/volcano/intro.mdx |
★标题 "Introduction" 需改为组件名 | 1 |
training_pipelines/training_guides/checkpointing-and-resuming.mdx |
train/how_to/checkpointing-and-resuming.mdx |
6 | |
training_pipelines/training_guides/fine-tune-and-pretrain-llms-on-ascend-npu.mdx |
train/how_to/fine-tune-and-pretrain-llms-on-ascend-npu.mdx |
4 | |
training_pipelines/training_guides/fine-tune-with-trainer-v2.mdx |
train/how_to/fine-tune-with-trainer-v2.mdx |
1 | |
training_pipelines/training_guides/fine-tuning-pipeline-with-mlflow-trustyai.mdx |
train/how_to/fine-tuning-pipeline-with-mlflow-trustyai.mdx |
22 | |
training_pipelines/training_guides/fine-tuning-using-notebooks.mdx |
train/how_to/fine-tuning-using-notebooks.mdx |
7 | |
training_pipelines/training_guides/gpu-slicing-with-dra.mdx |
train/how_to/gpu-slicing-with-dra.mdx |
4 | |
training_pipelines/training_guides/index.mdx |
train/how_to/index.mdx |
13 | |
training_pipelines/training_guides/kfp-execution-and-storage.mdx |
train/how_to/kfp-execution-and-storage.mdx |
||
training_pipelines/training_guides/kubeflow-trainer-quick-start.md |
train/how_to/kubeflow-trainer-quick-start.md |
||
training_pipelines/training_guides/preemptible-trainjobs-with-kueue.mdx |
train/how_to/preemptible-trainjobs-with-kueue.mdx |
4 | |
training_pipelines/training_guides/reusable-pipeline-components.mdx |
train/how_to/reusable-pipeline-components.mdx |
1 | |
training_pipelines/training_guides/training-hub-fine-tuning.mdx |
train/how_to/training-hub-fine-tuning.mdx |
2 | |
training_pipelines/training_guides/training-runtimes.mdx |
train/how_to/training-runtimes.mdx |
3 | |
training_pipelines/kueue/how_to/config_quotas.mdx |
train/quota_scheduling/config_quotas.mdx |
||
training_pipelines/kueue/how_to/fair_sharing.mdx |
train/quota_scheduling/fair_sharing.mdx |
||
training_pipelines/kueue/how_to/gang_scheduling.mdx |
train/quota_scheduling/gang_scheduling.mdx |
||
training_pipelines/kueue/how_to/npu_quota.mdx |
train/quota_scheduling/npu_quota.mdx |
||
training_pipelines/kueue/how_to/setup_rbac.mdx |
train/quota_scheduling/setup_rbac.mdx |
||
training_pipelines/kueue/how_to/using_cohorts.mdx |
train/quota_scheduling/using_cohorts.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
model_inference/envoy_ai_gateway/index.mdx |
deploy/components/envoy_ai_gateway/index.mdx |
||
model_inference/envoy_ai_gateway/install.mdx |
deploy/components/envoy_ai_gateway/install.mdx |
||
model_inference/envoy_ai_gateway/intro.mdx |
deploy/components/envoy_ai_gateway/intro.mdx |
★标题 "Introduction" 需改为组件名 | 7 |
model_inference/infernex_bridge/index.mdx |
deploy/components/infernex_bridge/index.mdx |
||
model_inference/infernex_bridge/install.mdx |
deploy/components/infernex_bridge/install.mdx |
3 | |
model_inference/infernex_bridge/intro.mdx |
deploy/components/infernex_bridge/intro.mdx |
★标题 "Introduction" 需改为组件名 | 1 |
model_inference/kserve/index.mdx |
deploy/components/kserve/index.mdx |
||
model_inference/kserve/install.mdx |
deploy/components/kserve/install.mdx |
2 | |
model_inference/kserve/intro.mdx |
deploy/components/kserve/intro.mdx |
★标题 "Introduction" 需改为组件名 | 1 |
model_inference/lws/index.mdx |
deploy/components/lws/index.mdx |
||
model_inference/lws/install.mdx |
deploy/components/lws/install.mdx |
||
model_inference/lws/intro.mdx |
deploy/components/lws/intro.mdx |
★标题 "Introduction" 需改为组件名 | 2 |
model_inference/envoy_ai_gateway/how_to/cost_management_chargeback.mdx |
deploy/gateway/cost_management_chargeback.mdx |
4 | |
model_inference/envoy_ai_gateway/how_to/external_provider_routing.mdx |
deploy/gateway/external_provider_routing.mdx |
3 | |
model_inference/envoy_ai_gateway/how_to/identity_authentication.mdx |
deploy/gateway/identity_authentication.mdx |
6 | |
model_inference/envoy_ai_gateway/how_to/token_rate_limiting.mdx |
deploy/gateway/token_rate_limiting.mdx |
6 | |
model_inference/envoy_ai_gateway/how_to/usage_metering.mdx |
deploy/gateway/usage_metering.mdx |
8 | |
training_pipelines/kueue/how_to/isvc.mdx |
deploy/how_to/isvc.mdx |
||
model_inference/inference_service/functions/index.mdx |
deploy/inference_service/functions/index.mdx |
||
model_inference/inference_service/functions/inference_service.mdx |
deploy/inference_service/functions/inference_service.mdx |
||
model_inference/inference_service/how_to/accurately_schedule.mdx |
deploy/inference_service/how_to/accurately_schedule.mdx |
||
model_inference/inference_service/how_to/autoscale_settings.mdx |
deploy/inference_service/how_to/autoscale_settings.mdx |
||
model_inference/inference_service/how_to/create_inference_service_cli.mdx |
deploy/inference_service/how_to/create_inference_service_cli.mdx |
||
model_inference/inference_service/how_to/custom_inference_runtime.mdx |
deploy/inference_service/how_to/custom_inference_runtime.mdx |
||
model_inference/inference_service/how_to/external_access_inference_service.mdx |
deploy/inference_service/how_to/external_access_inference_service.mdx |
||
model_inference/inference_service/how_to/index.mdx |
deploy/inference_service/how_to/index.mdx |
||
model_inference/inference_service/how_to/keda_autoscaling.mdx |
deploy/inference_service/how_to/keda_autoscaling.mdx |
||
model_inference/inference_service/how_to/using_modelcar.mdx |
deploy/inference_service/how_to/using_modelcar.mdx |
2 | |
model_inference/inference_service/how_to/vllm_expert_parallel.mdx |
deploy/inference_service/how_to/vllm_expert_parallel.mdx |
3 | |
model_inference/inference_service/how_to/vllm_speculative_decoding.mdx |
deploy/inference_service/how_to/vllm_speculative_decoding.mdx |
6 | |
model_inference/inference_service/index.mdx |
deploy/inference_service/index.mdx |
||
model_inference/inference_service/intro.mdx |
deploy/inference_service/intro.mdx |
||
model_inference/inference_service/trouble_shooting/index.mdx |
deploy/inference_service/trouble_shooting/index.mdx |
||
model_inference/inference_service/trouble_shooting/infer_timeout.mdx |
deploy/inference_service/trouble_shooting/infer_timeout.mdx |
||
model_inference/inference_service/trouble_shooting/pod_security_admission_violation.mdx |
deploy/inference_service/trouble_shooting/pod_security_admission_violation.mdx |
||
model_inference/maas/index.mdx |
deploy/maas/index.mdx |
||
model_inference/maas/intro.mdx |
deploy/maas/intro.mdx |
||
model_inference/llm_compressor/how_to/compressor_by_workbench.mdx |
deploy/model_compression/compressor_by_workbench.mdx |
4 | |
model_inference/llm_compressor/how_to/index.mdx |
deploy/model_compression/how_to/index.mdx |
||
model_inference/llm_compressor/index.mdx |
deploy/model_compression/index.mdx |
||
model_inference/llm_compressor/intro.mdx |
deploy/model_compression/intro.mdx |
||
model_inference/model_management/functions/index.mdx |
deploy/model_management/functions/index.mdx |
||
model_inference/model_management/functions/model_repository.mdx |
deploy/model_management/functions/model_repository.mdx |
1 | |
model_inference/model_management/functions/model_storage.mdx |
deploy/model_management/functions/model_storage.mdx |
5 | |
model_inference/model_management/how_to/index.mdx |
deploy/model_management/how_to/index.mdx |
||
model_inference/model_management/how_to/share_models.mdx |
deploy/model_management/how_to/share_models.mdx |
||
model_inference/model_management/index.mdx |
deploy/model_management/index.mdx |
||
model_inference/model_management/intro.mdx |
deploy/model_management/intro.mdx |
||
model_inference/overview/features.mdx |
deploy/overview/features.mdx |
||
model_inference/overview/index.mdx |
deploy/overview/index.mdx |
||
model_inference/overview/intro.mdx |
deploy/overview/intro.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
ai_applications/dify/overview/features.mdx |
ai_applications/components/dify/features.mdx |
打平 overview/ 子目录 | 1 |
ai_applications/dify/index.mdx |
ai_applications/components/dify/index.mdx |
||
ai_applications/dify/install.mdx |
ai_applications/components/dify/install.mdx |
1 | |
ai_applications/dify/overview/intro.mdx |
ai_applications/components/dify/intro.mdx |
打平 overview/ 子目录 ; ★标题 "Introduction" 需改为组件名 | 3 |
ai_applications/kagenti/index.mdx |
ai_applications/components/kagenti/index.mdx |
||
ai_applications/kagenti/install.mdx |
ai_applications/components/kagenti/install.mdx |
3 | |
ai_applications/kagenti/intro.mdx |
ai_applications/components/kagenti/intro.mdx |
★标题 "Introduction" 需改为组件名 | |
ai_applications/kagenti/security_architecture.mdx |
ai_applications/components/kagenti/security_architecture.mdx |
4 | |
ai_applications/llama_stack/overview/features.mdx |
ai_applications/components/llama_stack/features.mdx |
打平 overview/ 子目录 | |
ai_applications/llama_stack/index.mdx |
ai_applications/components/llama_stack/index.mdx |
||
ai_applications/llama_stack/install.mdx |
ai_applications/components/llama_stack/install.mdx |
2 | |
ai_applications/llama_stack/overview/intro.mdx |
ai_applications/components/llama_stack/intro.mdx |
打平 overview/ 子目录 ; ★标题 "Introduction" 需改为组件名 | |
ai_applications/llama_stack/quickstart.mdx |
ai_applications/components/llama_stack/quickstart.mdx |
2 | |
ai_applications/mcp_lifecycle_operator/index.mdx |
ai_applications/components/mcp_lifecycle_operator/index.mdx |
||
ai_applications/mcp_lifecycle_operator/install.mdx |
ai_applications/components/mcp_lifecycle_operator/install.mdx |
||
ai_applications/mcp_lifecycle_operator/intro.mdx |
ai_applications/components/mcp_lifecycle_operator/intro.mdx |
★标题 "Introduction" 需改为组件名 | 1 |
ai_applications/mcp_lifecycle_operator/quickstart.mdx |
ai_applications/components/mcp_lifecycle_operator/quickstart.mdx |
1 | |
ai_applications/kagenti/how_to/demo_advanced_secure_profile.mdx |
ai_applications/how_to/demo_advanced_secure_profile.mdx |
6 | |
ai_applications/kagenti/how_to/demo_with_secure_profile.mdx |
ai_applications/how_to/demo_with_secure_profile.mdx |
10 | |
ai_applications/kagenti/how_to/deploy_agent_with_agentruntime.mdx |
ai_applications/how_to/deploy_agent_with_agentruntime.mdx |
8 | |
ai_applications/kagenti/how_to/envoy_ai_gateway_mcp.mdx |
ai_applications/how_to/envoy_ai_gateway_mcp.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
ai_applications/trustyai/ai-guardrails.mdx |
evaluate_safety/ai_guardrails.mdx |
重命名（连字符→下划线） | 1 |
ai_applications/trustyai/index.mdx |
evaluate_safety/components/trustyai/index.mdx |
||
ai_applications/trustyai/install.mdx |
evaluate_safety/components/trustyai/install.mdx |
4 | |
ai_applications/trustyai/intro.mdx |
evaluate_safety/components/trustyai/intro.mdx |
★标题 "Introduction" 需改为组件名 | 5 |
ai_applications/trustyai/lm-eval.mdx |
evaluate_safety/lm_eval.mdx |
重命名（连字符→下划线） | 1 |
ai_applications/trustyai/nemo-guardrails.mdx |
evaluate_safety/nemo_guardrails.mdx |
重命名（连字符→下划线） | 1 |
ai_applications/ragas.mdx |
evaluate_safety/ragas.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
ai_applications/trustyai/tas.mdx |
monitor/bias_drift.mdx |
1 | |
training_pipelines/kueue/how_to/monitor_pending_workload.mdx |
monitor/how_to/monitor_pending_workload.mdx |
||
monitoring_ops/index.mdx |
monitor/index.mdx |
||
monitoring_ops/logging_tracing/functions/index.mdx |
monitor/logging_tracing/functions/index.mdx |
||
monitoring_ops/logging_tracing/functions/logging.mdx |
monitor/logging_tracing/functions/logging.mdx |
||
monitoring_ops/logging_tracing/index.mdx |
monitor/logging_tracing/index.mdx |
||
monitoring_ops/logging_tracing/intro.mdx |
monitor/logging_tracing/intro.mdx |
||
monitoring_ops/overview/features.mdx |
monitor/overview/features.mdx |
||
monitoring_ops/overview/index.mdx |
monitor/overview/index.mdx |
||
monitoring_ops/overview/intro.mdx |
monitor/overview/intro.mdx |
||
monitoring_ops/resource_monitoring/functions/index.mdx |
monitor/resource_monitoring/functions/index.mdx |
||
monitoring_ops/resource_monitoring/functions/resource_monitoring.mdx |
monitor/resource_monitoring/functions/resource_monitoring.mdx |
||
monitoring_ops/resource_monitoring/how_to/add_monitor_dashboard.mdx |
monitor/resource_monitoring/how_to/add_monitor_dashboard.mdx |
||
monitoring_ops/resource_monitoring/how_to/index.mdx |
monitor/resource_monitoring/how_to/index.mdx |
||
monitoring_ops/resource_monitoring/index.mdx |
monitor/resource_monitoring/index.mdx |
||
monitoring_ops/resource_monitoring/intro.mdx |
monitor/resource_monitoring/intro.mdx |
||
monitoring_ops/resource_monitoring/troubleshooting/index.mdx |
monitor/resource_monitoring/troubleshooting/index.mdx |
||
monitoring_ops/resource_monitoring/troubleshooting/monitor_dashboard_loading_stuck.mdx |
monitor/resource_monitoring/troubleshooting/monitor_dashboard_loading_stuck.mdx |
| 现有路径 | 新路径 | 备注 | 需改链接 |
|---|---|---|---|
apis/index.mdx |
apis/index.mdx |
||
apis/intro.mdx |
apis/intro.mdx |
||
apis/kubernetes_apis/index.mdx |
apis/kubernetes_apis/index.mdx |
||
apis/kubernetes_apis/kubeflow.org/index.mdx |
apis/kubernetes_apis/kubeflow.org/index.mdx |
||
apis/kubernetes_apis/kubeflow.org/workspacekinds.mdx |
apis/kubernetes_apis/kubeflow.org/workspacekinds.mdx |
||
apis/kubernetes_apis/kubeflow.org/workspaces.mdx |
apis/kubernetes_apis/kubeflow.org/workspaces.mdx |
||
apis/kubernetes_apis/manage/amlnamespaces.mdx |
apis/kubernetes_apis/manage/amlnamespaces.mdx |
||
apis/kubernetes_apis/manage/index.mdx |
apis/kubernetes_apis/manage/index.mdx |
||
apis/kubernetes_apis/operator/amlclusters.mdx |
apis/kubernetes_apis/operator/amlclusters.mdx |
||
apis/kubernetes_apis/operator/index.mdx |
apis/kubernetes_apis/operator/index.mdx |
||
apis/kubernetes_apis/serving.kserve.io/clusterservingruntimes.mdx |
apis/kubernetes_apis/serving.kserve.io/clusterservingruntimes.mdx |
||
apis/kubernetes_apis/serving.kserve.io/index.mdx |
apis/kubernetes_apis/serving.kserve.io/index.mdx |
||
apis/kubernetes_apis/serving.kserve.io/inferenceservices.mdx |
apis/kubernetes_apis/serving.kserve.io/inferenceservices.mdx |
| 路径 | 原因 |
|---|---|
ai_applications/dify/overview/index.mdx |
删除（overview/ 打平后冗余） |
ai_applications/index.mdx |
删除（pillar 解散） |
ai_applications/kagenti/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
ai_applications/llama_stack/overview/index.mdx |
删除（overview/ 打平后冗余） |
data_experiments/index.mdx |
删除（pillar 解散） |
data_experiments/label_studio/overview/index.mdx |
删除（overview/ 打平后冗余） |
data_experiments/mlflow/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
infrastructure_management/index.mdx |
删除（pillar 解散） |
learn/index.mdx |
删除（pillar 解散） |
model_inference/envoy_ai_gateway/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
model_inference/index.mdx |
删除（pillar 解散） |
training_pipelines/data_science_pipelines/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
training_pipelines/index.mdx |
删除（pillar 解散） |
training_pipelines/kubeflow/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
training_pipelines/kuberay/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
training_pipelines/kueue/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
training_pipelines/spark_operator/how_to/index.mdx |
删除（how_to/ 下的页面已迁往场景章节，该分组页冗余） |
19 个组件的 intro.mdx 标题目前一律是 # Introduction，并列在 components/ 下无法区分，
需改为组件名（例如 # Alauda Build of Kueue）。
dify / llama_stack / label_studio 的 overview/ 子目录在组件移入 components/ 后会导致 5 层嵌套，
需打平为组件目录下的平级页面。
章节级 7 篇（plan administer develop train deploy ai_applications evaluate_safety）——
含 h1、<Overview />，以及一段**「本章涉及的组件」导语**，链接到住在其它章节的依赖组件。
这是替代占位页的机制。
另有 5 个章节的 index.mdx 由现有目录继承（overview installation upgrade apis monitor），
其中 monitor/index.mdx 的 h1 需从 Monitoring & Ops 改为 Monitor。
分组级 10 篇 —— 按规范每个目录都需要 index.mdx：
plan/、evaluate_safety/、develop/how_to/、develop/experiment_tracking/、
train/quota_scheduling/、deploy/how_to/、deploy/gateway/、administer/how_to/、
ai_applications/how_to/、monitor/how_to/。
新结构最深 4 层（章节 > 模块 > 分组 > 页面），与现状持平，0 处超标。 层数分布：2 层 26 篇、3 层 84 篇、4 层 131 篇。
全仓 51 篇 how_to 中仅 30 篇有 Prerequisites，且部分为纯文本未加链接
（如 kueue/how_to/isvc.mdx 的「You have installed the Alauda Build of Kueue」）。
迁移后需补齐并统一加上指向组件安装页的链接。
| 场景 | 依赖的组件 | 组件所在章节 | 现有引用数 |
|---|---|---|---|
| Train | Kueue | Train（本章） | 7 → 归零 |
| Develop | Envoy AI Gateway | Deploy | 5 |
| Train | MLflow | Develop | 4 |
| Overview | Dify | Build AI Apps | 3 |
| Upgrade | KServe | Deploy | 2 |
| Administer | MLflow | Develop | 2 |
| Safety | TrustyAI | Evaluate & Safety（本章） | 2 → 归零 |
| Administer | Kagenti | Build AI Apps | 1 |
| Install | KServe | Deploy | 1 |
| Monitor | TrustyAI | Evaluate & Safety | 1 |
- Monitor 独立成章 — 不并入 Administer。监控能力横跨使用者与管理员两个视图
（
resource_monitoring/how_to/add_monitor_dashboard.mdx明确要求切换到 Administrator View， 而logging.mdx/resource_monitoring.mdx面向推理服务使用者），跨视角的能力应独立为场景。 - Feast 归 Develop — 面向数据科学家的特征库，与 MLflow / Label Studio 同章。 （RHOAI 将 Feature Store 置于 Administer，此处按国内团队的使用方式选择 Develop。）
- Kueue 归 Train — 连同配额、公平共享、Cohort、RBAC、NPU 配额 5 篇治理页一并移入
train/quota_scheduling/。 理由：引用 Kueue 安装页的 7 处全部来自训练场景；这 5 篇本质是「为训练做的资源治理」而非平台级管理。 该选择同时消除了最重的一组跨场景依赖（Train→Kueue 7 处归零）。 - 不设占位页 — 次要场景通过任务页 Prerequisites 的内联链接 + 章节导语指向组件， 不新增仅含跳转的导航条目（RHOAI 全套文档亦无此类页面）。
- 不回合
release-2.7— 本次重构使全部 URL 变更，仅进入 2.8。已发布的 2.7/2.6 由各自 release 分支构建，不受影响；若需回合 2.7 则本方案不可行。 - 版本切换器 — 从 2.7 切到 2.8 时路径失效，需确认切换器对不存在的路径有兜底。
- llms.txt 需重新生成 —
grouping_base_path: docs/en按一级目录分组，重构后现有分组全部失效。 - 这是连续第二次重排 — PR #297 刚合并。翻译 sourceSHA、llms.txt、外部引用的成本会叠加， 建议本方案一次到位，不再分期改动目录结构。
第一版结构保留了旧模块模板的 how_to/ / functions/ 分组，评审中暴露三个问题：
- 章节级 How To 与同级模块语义重叠。
develop/how_to/的 10 篇任务页与同级的 Workbench、Connections、Experiment Tracking 讲的是同一批工作，读者无法判断 "Use Kubeflow Notebooks" 应该去 Workbench 还是去 How To。administer/、deploy/、monitor/、ai_applications/也各有一个同样的桶，共 5 个。 - "How To" 不携带信息。 该标签只说明体裁不说明内容；RHOAI 的导航节点一律以对象或任务命名 （Working with connections、Managing resources…），不存在按体裁分组的节点。
- 单页包装目录。
deploy/model_compression/how_to/只有 index 没有任何页面；另有 6 个how_to/functions/overview/troubleshooting/目录只包 1 篇文章。
- 与模块并列的章节级
how_to/一律拆散，按主题并入既有模块，或新建以任务命名的分组。 - 模块内的分组：≤4 篇直接打平到模块下；>4 篇保留分组，但重命名为描述性标签。
- 只含 1 篇文章的包装目录一律打平；空目录删除。
- 全站不再出现
How To导航标签，目录名同步去掉how_to（URL 与标签一致）。
| 章节 | 原位置 | 新位置 |
|---|---|---|
| Administer | administer/how_to/mlflow_workspaces.mdx |
administer/multi_tenant/（标题改为 MLflow Workspaces and Access Control） |
| Administer | administer/how_to/{install_secure_dependencies,enable_secure_profile}.mdx |
administer/secure_profile/（新建分组） |
| Administer | administer/hardware_profile/{functions,how_to}/ |
打平到 administer/hardware_profile/ |
| Administer | administer/multi_tenant/functions/ |
打平到 administer/multi_tenant/ |
| Develop | develop/how_to/{notebooks,volumes-kserve,tensorboards,upload_models_using_notebook}.mdx |
develop/workbench/ |
| Develop | develop/workbench/{overview,how_to}/、develop/connections/{overview,how_to}/ |
各自打平到模块下 |
| Develop | develop/how_to/{create_dspa,pipelines,tekton}.mdx + workbench/how_to/run-kubeflow-pipelines-with-elyra.mdx + train/how_to/{reusable-pipeline-components,kfp-execution-and-storage}.mdx |
develop/pipelines/（新建分组，pipelines.mdx 更名 kubeflow_pipelines.mdx） |
| Develop | develop/how_to/{codeflare-sdk-tutorial,run_spark_application}.mdx |
develop/distributed_workloads/（新建分组） |
| Develop | develop/how_to/model-registry.mdx |
develop/model_registry.mdx（章节级单页） |
| Train | train/how_to/ |
train/guides/（标签本已是 Training Guides，仅对齐目录名） |
| Deploy | deploy/how_to/isvc.mdx |
deploy/inference_service/guides/kueue_scheduling.mdx（标题改为 Schedule Inference Services with Kueue） |
| Deploy | deploy/inference_service/how_to/（9 篇） |
deploy/inference_service/guides/（10 篇，标签 Guides） |
| Deploy | deploy/inference_service/functions/inference_service.mdx |
打平到模块下（标题改为 Managing Inference Services） |
| Deploy | deploy/inference_service/trouble_shooting/ |
deploy/inference_service/troubleshooting/ |
| Deploy | deploy/model_management/{functions,how_to}/ |
打平到 deploy/model_management/ |
| Deploy | deploy/maas/（仅 1 篇 intro） |
deploy/maas.mdx |
| Deploy | deploy/model_compression/how_to/ |
删除（空目录） |
| Build AI Apps | ai_applications/how_to/（4 篇） |
章节根，与 Evaluate & Safety 一致 |
| Monitor | monitor/how_to/monitor_pending_workload.mdx |
monitor/resource_monitoring/ |
| Monitor | monitor/{logging_tracing,resource_monitoring}/{functions,how_to,troubleshooting}/ |
各自打平到模块下（resource_monitoring.mdx 标题改为 Monitoring Metrics and Views） |
两篇 KFP 通用页（Reusable Pipeline Components、Execution and Storage Behavior）从 Train 移到 Develop > Pipelines：它们讲的是 KFP 本身的能力与行为，不是训练方法；Train 的索引表保留指向它们的跨章链接。
- 删除 21 个 index.mdx，新建 3 个（
develop/pipelines/、develop/distributed_workloads/、administer/secure_profile/），总页数 252 → 234。 - 二级 How To 节点 5 个 → 0；全站
how_to路径 113 处 → 0（ExternalSiteLink指向 ACP 站点的除外）。 - 相对链接、GitHub raw/tree URL、3 个 e2e 脚本的 assets 路径已同步重写，校验 0 处失效链接、 0 个目录缺 index.mdx、每个目录的 weight 均为 10 步长无重复。
- §4.3 中"分组级 index.mdx"的清单已被本节取代。
按"每篇文章的内容 vs 它的位置"复核 §8 结果后修正的 5 项：
| # | 问题 | 处理 |
|---|---|---|
| 1 | doom 的路由是 **/*.md{,x}，但 §8 的 weight 重排只扫了 .mdx，把 plan/supported_configurations.md（weight 10）和 train/guides/kubeflow-trainer-quick-start.md（weight 10）撞了 |
Plan 恢复 20/30/40，Training Guides 恢复 20…100。这两个目录"10 的空缺"本来就是那个 .md 页 |
| 2 | develop/pipelines/tekton.mdx 正文讲的是"用 Kueue 的配额管住 Tekton PipelineRun"，标题 Integrate with Alauda DevOps Pipelines 让它看起来像 KFP 页 |
改名 Schedule Alauda DevOps Pipelines with Kueue，与同源的 kueue_scheduling.mdx 对齐 |
| 3 | Kubeflow Notebooks 是 Workbench 的替代品（notebooks.mdx 自己就这么写），放进 Workbench 模块等于说它是用 Workbench 的方法 |
拆出平级分组 develop/kubeflow_notebooks/（Notebooks、Volumes、Tensorboards），index 说明二者关系与 Kubeflow operator 前置 |
| 5 | upload_models_using_notebook.mdx 是"往模型仓库传模型"的任务（入链 Deploy 3 + Train 3 + Develop 0），被按"用了 notebook"归到 Workbench |
移到 deploy/model_management/，排在 Model Repository 之后 |
| 9 | 两个 index 的 H1 与导航标签不符：Validated Models → # Inference Guide，Device Options → # Device Management |
H1 与标签对齐，并修正一处引用旧名的链接文字 |
仍待决策（本轮未改）：volumes-kserve.mdx 后半段其实是 KServe 模型部署，与标题不符，建议拆页；
deploy/overview/ 的 Introduction 是 Model Management 与 Inference Service 两篇 intro 的逐字拼接，
monitor/overview/ 同形——场景化导航里章节 index 即 Overview，这两个 Overview 模块建议删除。
volumes-kserve.mdx 拆页。 该页前半是 Kubeflow Volumes（建卷、管卷、挂到 Notebook），
后半 4 节是从 Kubeflow 面板的 KServe Endpoints UI 部署模型，导航标签 Use Kubeflow Volumes
完全盖住了后半段。拆为：
develop/kubeflow_notebooks/volumes.mdx—— 只留卷相关内容，末尾指向下一页；deploy/inference_service/guides/kubeflow_kserve_endpoints.mdx—— Deploy Inference Services from the Kubeflow Dashboard，补齐 Introduction / Prerequisites / Verification， 并说明它创建的是与 CLI、Alauda AI 控制台同一个InferenceService对象，只是入口不同。 排在 Guides 的 CLI 创建页之后。
删除 deploy/overview/ 与 monitor/overview/。 场景化导航里章节 index.mdx 就是 Overview，
再挂一个 Overview 模块是旧模块模板的残留，且内容是重复的：
| 页面 | 与谁重复 |
|---|---|
deploy/overview/intro.mdx |
正文 = model_management/intro.mdx + inference_service/intro.mdx 逐字拼接 |
deploy/overview/features.mdx |
model_repository.mdx 与 inference_service.mdx 的 Advantages / Core Features 的弱化版 |
monitor/overview/intro.mdx |
monitor/index.mdx 章节导语 |
monitor/overview/features.mdx |
logging.mdx 的 Core Features 与 resource_monitoring.mdx 的 Main Features 的弱化版 |
唯一不重复的一条是 monitor/overview/intro.mdx 里的「Hami GPU 面板需 1.4+」，已移入
resource_monitoring/intro.mdx 的 Usage Limitations 并补上指向 Hami 页的链接；
Monitor 章节导语补了一句全生命周期可观测性的表述。两个 Overview 模块无任何入链。
monitor/bias_drift.mdx 的位置本身成立——它接线上流量、跑定时任务、出口是 Prometheus 面板，
与同章的 Logging、Resource Monitoring 同属"服务跑起来之后持续看"，而 LM-Eval 是一次性评测。
但有两处不一致：
- 导航标签用的是产品名（
TrustyAI Service (TAS)），同级兄弟却按能力命名（Logging & Tracing、 Resource Monitoring）。改为 Bias and Drift Monitoring（与文件名bias_drift.mdx一致）， 产品全称保留在正文首段。 - 页面前 40% 是组件部署（Prerequisites → Deploy TrustyAIService → DATABASE / PVC → Verify），
而 TrustyAI 组件页在 Evaluate & Safety。这半截切给
evaluate_safety/components/trustyai/deploy_tas.mdx（Deploy TrustyAI Service，weight 30）， Monitor 页只留 Access the TAS API、数据摄取、drift / bias 指标与 Prometheus 部分， Prerequisites 改为指向新页。
两页互链，components/trustyai/{intro,install}.mdx 与 Monitor 章节导语的引用同步更新。
组件仍留在 Evaluate & Safety：TrustyAI operator 的 4 个消费方里 3 个（LM-Eval、FMS Guardrails、
NeMo Guardrails）在该章。
plan/glossary.mdx → overview/glossary.mdx（weight 50，排在 Release Notes 之后）。
§3 的第一版映射把原 learn/ 整章折进 Plan，glossary 是顺带过去的。但 Plan 的定位是
"装机前的选型决策"——章节导语只介绍 supported configurations、validated models、
device options 三块，从未引用 glossary，它在这一章里是无入链的孤页。术语表是概念参考，
与 overview/ 的 Intro、Architecture 同层。
与 RHOAI 3.x 对照也支持这一归位：其 Plan 章只有 Choose a validated model for reliable serving
与 Prepare your platform and hardware for Red Hat AI 两篇，对应我们的 validated_models/ 与
device_options/ + supported_configurations.md；RHOAI 顶层另有的 Learn 章是尾部的外部学习
资源聚合位（Supported Configurations 门户页、Red Hat AI Foundations、Red Hat AI learning hub，
三条全是站外链接），不是"规划"章的同义改名。Alauda AI 没有对应的外部学习站点，因此不新建
Learn 章，Plan 保持现名。
全仓无指向该页的相对链接，移动无需改链；llms.txt 与 llmstxt-state.json 的条目同步移入
overview 分组。