了解由21Vianet區域運營的Azure Government、Azure 德國及Microsoft Azure中可用的Azure Machine Learning功能。
在全球 Azure 區域列表中,除了公共雲區域外,還有數個服務特定市場的區域。 例如 Azure Government 與由世紀互聯提供的 Microsoft Azure 區域。 除了公共雲區域外,Azure Machine Learning 已部署於以下區域:
- Azure 政府地區美國-亞利桑那及美國-維吉尼亞。
- Azure由21Vianet區域中國東2運營。
本文件其餘部分提供 Azure Machine Learning 在這些區域可用的功能,以及區域特定的使用資訊。
Azure Government
USGOV Virginia 與 Arizona 的模型目錄
| 模型名稱 | MaaP 推斷 | 已啟用批次 | Standard 部署 (無伺服器計算推斷) | Standard 部署 (微調) |
|---|---|---|---|---|
| ALLaM-2-7b-instruct | maap-inference | |||
| Aurora | maap-inference | |||
| AutoML-Image-Classification | maap-inference | batch-enabled | ||
| AutoML-Image-Instance-Segmentation | maap-inference | batch-enabled | ||
| AutoML-Image-Object-Detection | maap-inference | batch-enabled | ||
| bert-base-cased | maap-inference | batch-enabled | ||
| bert-base-uncased | maap-inference | batch-enabled | ||
| bert-large-cased | maap-inference | batch-enabled | ||
| bert-large-uncased | maap-inference | batch-enabled | ||
| BiomedCLIP-PubMedBERT_256-vit_base_patch16_224 | maap-inference | batch-enabled | ||
| bytetrack_yolox_x_crowdhuman_mot17-private-half | maap-inference | batch-enabled | ||
| camembert-base | maap-inference | batch-enabled | ||
| CodeLlama-13b-hf | maap-inference | batch-enabled | ||
| CodeLlama-13b-Instruct-hf | maap-inference | batch-enabled | ||
| CodeLlama-13b-Python-hf | maap-inference | batch-enabled | ||
| CodeLlama-34b-hf | maap-inference | batch-enabled | ||
| CodeLlama-34b-Instruct-hf | maap-inference | batch-enabled | ||
| CodeLlama-34b-Python-hf | maap-inference | batch-enabled | ||
| CodeLlama-70b-hf | maap-inference | |||
| CodeLlama-70b-Instruct-hf | maap-inference | |||
| CodeLlama-70b-Python-hf | maap-inference | |||
| CodeLlama-7b-hf | maap-inference | batch-enabled | ||
| CodeLlama-7b-Instruct-hf | maap-inference | batch-enabled | ||
| CodeLlama-7b-Python-hf | maap-inference | batch-enabled | ||
| compvis-stable-diffusion-v1-4 | maap-inference | batch-enabled | ||
| CxrReportGen maap-inference | batch-enabled | |||
| databricks-dolly-v2-12b | maap-inference | batch-enabled | ||
| Deci-DeciCoder-1b | maap-inference | batch-enabled | ||
| deci-decidiffusion-v1-0 | maap-inference | batch-enabled | ||
| Deci-DeciLM-7B | maap-inference | batch-enabled | ||
| Deci-DeciLM-7B-instruct | maap-inference | batch-enabled | ||
| deepset-minilm-uncased-squad2 | maap-inference | batch-enabled | ||
| deepset-roberta-base-squad2 | maap-inference | batch-enabled | ||
| distilbert-base-cased | maap-inference | batch-enabled | ||
| distilbert-base-cased-distilled-squad | maap-inference | batch-enabled | ||
| distilbert-base-uncased | maap-inference | batch-enabled | ||
| distilbert-base-uncased-distilled-squad | maap-inference | batch-enabled | ||
| distilbert-base-uncased-finetuned-sst-2-english | maap-inference | batch-enabled | ||
| distilgpt2 | maap-inference | batch-enabled | ||
| distilroberta-base | maap-inference | batch-enabled | ||
| facebook-bart-large-cnn | maap-inference | batch-enabled | ||
| facebook-deit-base-patch16-224 | maap-inference | batch-enabled | ||
| facebook-dinov2-base-imagenet1k-1-layer | maap-inference | batch-enabled | ||
| Facebook-DinoV2-Image-Embeddings-ViT-Base | maap-inference | batch-enabled | ||
| Facebook-DinoV2-圖像嵌入-ViT-Giant | maap-inference | batch-enabled | ||
| facebook-sam-vit-base | maap-inference | batch-enabled | ||
| facebook-sam-vit-huge | maap-inference | batch-enabled | ||
| facebook-sam-vit-large | maap-inference | batch-enabled | ||
| financial-reports-analysis | maap-inference | batch-enabled | ||
| finiteautomata-bertweet-base-sentiment-analysis | maap-inference | batch-enabled | ||
| google-vit-base-patch16-224 | maap-inference | batch-enabled | ||
| Jean-Baptiste-camembert-ner | maap-inference | batch-enabled | ||
| Llama-2-13b | maap-inference | batch-enabled | maas-inference | maas-finetuning |
| Llama-2-13b-chat | maap-inference | maas-inference | ||
| Llama-2-70b | maap-inference | batch-enabled | maas-inference | maas-finetuning |
| Llama-2-70b-chat | maap-inference | maas-inference | ||
| Llama-2-7b maap-inference | batch-enabled | maas-inference | maas-finetuning | |
| Llama-2-7b-chat | maap-inference | maas-inference | ||
| Llama-3.2-11B-Vision-Instruct | maap-inference | maas-inference | ||
| Llama-3.2-1B | maap-inference | |||
| Llama-3.2-1B-Instruct | maap-inference | |||
| Llama-3.2-3B | maap-inference | |||
| Llama-3.2-3B-Instruct | maap-inference | |||
| Llama-3.2-90B-Vision-Instruct | maap-inference | maas-inference | ||
| Llama-Guard-3-11B-Vision | maap-inference | |||
| Llama-Guard-3-1B | maap-inference | |||
| Llama-Guard-3-8B | maap-inference | |||
| MatterSim | maap-inference | |||
| MedImageInsight | maap-inference | batch-enabled | ||
| MedImageParse | maap-inference | batch-enabled | ||
| Meta-Llama-3-70B | maap-inference | |||
| Meta-Llama-3-70B-Instruct | maap-inference | maas-inference | ||
| Meta-Llama-3-8B maap-inference | ||||
| Meta-Llama-3-8B-Instruct | maap-inference | maas-inference | ||
| Meta-Llama-3.1-70B | maap-inference | |||
| Meta-Llama-3.1-70B-Instruct | maap-inference | maas-inference | maas-finetuning | |
| Meta-Llama-3.1-8B | maap-inference | |||
| Meta-Llama-3.1-8B-Instruct | maap-inference | maas-inference | maas-finetuning | |
| microsoft-beit-base-patch16-224-pt22k-ft22k | maap-inference | batch-enabled | ||
| microsoft-deberta-base | maap-inference | batch-enabled | ||
| microsoft-deberta-base-mnli | maap-inference | batch-enabled | ||
| microsoft-deberta-large | maap-inference | batch-enabled | ||
| microsoft-deberta-large-mnli | maap-inference | batch-enabled | ||
| microsoft-deberta-xlarge | maap-inference | batch-enabled | ||
| microsoft-llava-med-v1.5-mistral-7b | maap-inference | batch-enabled | ||
| microsoft-rad-dino | maap-inference | batch-enabled | ||
| microsoft-swinv2-base-patch4-window12-192-22k | maap-inference | batch-enabled | ||
| mistral-community-Mixtral-8x22B-v0-1 | maap-inference | |||
| mistralai-Mistral-7B-Instruct-v0-2 | maap-inference | |||
| mistralai-Mistral-7B-Instruct-v01 (英文) | maap-inference | |||
| mistralai-Mistral-7B-v01 maap-inference | batch-enabled | |||
| mistralai-Mixtral-8x22B-Instruct-v0-1 | maap-inference | |||
| mistralai-Mixtral-8x22B-v0-1 | maap-inference | |||
| mistralai-Mixtral-8x7B-Instruct-v01 | maap-inference | |||
| mistralai-Mixtral-8x7B-v01 | maap-inference | |||
| mmd-3x-deformable-detr_refine_twostage_r50_16xb2-50e_coco | maap-inference | batch-enabled | ||
| mmd-3x-mask-rcnn_swin-t-p4-w7_fpn_1x_coco | maap-inference | batch-enabled | ||
| mmd-3x-sparse-rcnn_r101_fpn_300-proposals_crop-ms-480-800-3x_coco | maap-inference | batch-enabled | ||
| mmd-3x-sparse-rcnn_r50_fpn_300-proposals_crop-ms-480-800-3x_coco | maap-inference | batch-enabled | ||
| mmd-3x-vfnet_r50-mdconv-c3-c5_fpn_ms-2x_coco | maap-inference | batch-enabled | ||
| mmd-3x-vfnet_x101-64x4d-mdconv-c3-c5_fpn_ms-2x_coco | maap-inference | batch-enabled | ||
| mmd-3x-yolof_r50_c5_8x8_1x_coco | maap-inference | batch-enabled | ||
| mmeft | maap-inference | batch-enabled | ||
| Muse | maap-inference | |||
| ocsort_yolox_x_crowdhuman_mot17-private-half | maap-inference | batch-enabled | ||
| Phi-3-medium 指令 (128k) | maap-inference | maas-inference | maas-finetuning | |
| Phi-3-medium 指令 (4k) | maap-inference | maas-inference | maas-finetuning | |
| Phi-3-mini 指令 (128k) | maap-inference | maas-inference | maas-finetuning | |
| Phi-3-mini 指令 (4k) | maap-inference | maas-inference | maas-finetuning | |
| Phi-3-small 指令 (128k) | maap-inference | maas-inference | ||
| Phi-3-small 指令 (8k) | maap-inference | maas-inference | ||
| Phi-3-vision-128k-instruct | maap-inference | |||
| Phi-3.5-mini 指令 (128k) | maap-inference | maas-inference | maas-finetuning | |
| Phi-3.5-MoE 指令 (128k) | maap-inference | maas-inference | maas-finetuning | |
| Phi-3.5-vision 指令 (128k) | maap-inference | maas-inference | ||
| Phi-4 maap-推論 | maas-inference | |||
| 稜鏡 maap 推論 | batch-enabled | |||
| projecte-aina-aguila-7b | maap-inference | |||
| projecte-aina-FLOR-1-3B | maap-inference | |||
| projecte-aina-FLOR-1-3B-Instructed | maap-inference | |||
| projecte-aina-FLOR-6-3B | maap-inference | |||
| projecte-aina-FLOR-6-3B-Instructed | maap-inference | |||
| Prompt-Guard-86M | maap-inference | |||
| Prov-GigaPath | maap-inference | batch-enabled | ||
| roberta-base | maap-inference | batch-enabled | ||
| roberta-base-openai-detector | maap-inference | batch-enabled | ||
| roberta-large | maap-inference | batch-enabled | ||
| roberta-large-mnli | maap-inference | batch-enabled | ||
| roberta-large-openai-detector | maap-inference | batch-enabled | ||
| runwayml-stable-diffusion-inpainting | maap-inference | batch-enabled | ||
| runwayml-stable-diffusion-v1-5 | maap-inference | batch-enabled | ||
| Salesforce-BLIP-2-opt-2-7b-image-to-text | maap-inference | batch-enabled | ||
| Salesforce-BLIP-2-opt-2-7b-vqa | maap-inference | batch-enabled | ||
| Salesforce-BLIP-image-captioning-base | maap-inference | batch-enabled | ||
| Salesforce-BLIP-vqa-base | maap-inference | batch-enabled | ||
| snowflake-arctic-base | maap-inference | |||
| snowflake-arctic-instruct | maap-inference | |||
| sshleifer-distilbart-cnn-12-6 | maap-inference | batch-enabled | ||
| stabilityai-stable-diffusion-2-1 | maap-inference | batch-enabled | ||
| stabilityai-stable-diffusion-2-inpainting | maap-inference | batch-enabled | ||
| stabilityai-stable-diffusion-xl-base-1-0 | maap-inference | batch-enabled | ||
| stabilityai-stable-diffusion-xl-refiner-1-0 | maap-inference | batch-enabled | ||
| supply-chain-trade-regulations | maap-inference | batch-enabled | ||
| t5-base | maap-inference | batch-enabled | ||
| t5-large | maap-inference | batch-enabled | ||
| t5-small | maap-inference | batch-enabled | ||
| TamGen | maap-inference | |||
| tiiuae-falcon-40b | maap-inference | batch-enabled | ||
| tiiuae-falcon-7b | maap-inference | batch-enabled | ||
| Virchow | maap-inference | batch-enabled | ||
| Virchow2 | maap-inference | batch-enabled |
USGOV Virginia 與 Arizona 中模型目錄的 Azure OpenAI 模型
欲了解更多有關 Azure OpenAI 和 Azure Government 中由 Azure 直接銷售的所有 Foundry 模型的屬性,請參見 Azure Government 中由 Azure 直接銷售的模型。
Azure Machine Learning Features
| 功能 | 公用雲端狀態 | US-Virginia | US-Arizona |
|---|---|---|---|
| 自動化機器學習 | |||
| 在 Notebooks 中建立和執行實驗 | GA | 是 | 是 |
| 在 Studio Web 體驗中建立和執行實驗 | 公開預覽 | 是 | 是 |
| 業界領先的預測功能 | GA | 是 | 是 |
| 深度學習和其他進階學習者的支援 | GA | 是 | 是 |
| 大型資料支援 (最高 100 GB) | 公開預覽 | 是 | 是 |
| Azure Databricks 整合 | GA | 否 | 否 |
| SQL、Azure Cosmos DB 與 HDInsight 整合 | GA | 是 | 是 |
| 機器學習流程 | |||
| 使用 Azure Machine Learning SDK 建立、執行並發布管線 | GA | 是 | 是 |
| 使用 Azure Machine Learning SDK 建立管道端點 | GA | 是 | 是 |
| 使用 Azure Machine Learning SDK 建立、編輯及刪除管線的排程執行 | GA | 是* | 是* |
| 在 Studio 中檢視管線執行詳細資料 | GA | 是 | 是 |
| 在 Azure Machine Learning 設計工具 中建立、執行、視覺化及發佈管道 | GA | 是 | 是 |
| Azure Databricks 與機器學習管線的整合 | GA | 否 | 否 |
| 在 Azure Machine Learning 設計器中建立管線端點 | GA | 是 | 是 |
| 整合式筆記本 | |||
| 工作區筆記本和檔案共用 | GA | 是 | 是 |
| R 與 Python 支援 | GA | 是 | 是 |
| 虛擬網路支援 | GA | 是 | 是 |
| 計算執行個體 | |||
| 整合式筆記本的受控計算執行個體 | GA | 是 | 是 |
| Jupyter、JupyterLab 整合 | GA | 是 | 是 |
| 虛擬網路 (VNet) 支援 | GA | 是 | 是 |
| 設定 Apache Spark 集區以執行資料整理 | 公開預覽 | 否 | 否 |
| SDK 支援 | |||
| Python SDK support | GA | 是 | 是 |
| 安全性 | |||
| 受控虛擬網路支援 | 預覽 | 預覽 | 預覽 |
| 虛擬網路(VNet)訓練支援 | GA | 是 | 是 |
| 虛擬網路(VNet)推論支援 | GA | 是 | 是 |
| 評分端點驗證 | 公開預覽 | 是 | 是 |
| 工作區私用端點 | GA | GA | GA |
| VNet 後方的 ACI | 公開預覽 | 否 | 否 |
| VNet 後方的 ACR | GA | 是 | 是 |
| AKS 叢集的私用 IP | 公開預覽 | 否 | 否 |
| 受控線上端點的網路隔離 | GA | 否 | 否 |
| 計算 | |||
| 跨工作區的配額管理 | GA | 是 | 是 |
| Kubernetes 計算 | GA | 否 | 否 |
| 機器學習的資料 | |||
| 從 SDK 建立、檢視或編輯資料集與資料存放區 | GA | 是 | 是 |
| 從 UI 建立、檢視或編輯資料集與資料存放區 | GA | 是 | 是 |
| 從 SDK 檢視、編輯或刪除資料集漂移監視器 | 公開預覽 | 是 | 是 |
| 從 UI 檢視、編輯或刪除資料集漂移監視器 | 公開預覽 | 是 | 是 |
| 機器學習生命週期 | |||
| 模型剖析 (SDK/CLI v1) | GA | 是 | PARTIAL |
| The Azure Machine Learning CLI v1 | GA | 是 | 是 |
| Visual Studio Code 整合 | 公開預覽 | 否 | 否 |
| 事件方格整合 | 公開預覽 | 否 | 否 |
| 將Azure 串流分析整合到Azure Machine Learning | 公開預覽 | 否 | 否 |
| 標註影像與文字 | |||
| 標註項目管理入口網站 | GA | 是 | 是 |
| Labeler 入口網站 | GA | 是 | 是 |
| 使用私人工作區加標籤 | GA | 是 | 是 |
| ML 輔助標註 (影像分類與物件偵測) | 公開預覽 | 是 | 是 |
| 負責任的 ML | |||
| UI 中的可解釋性 | 公開預覽 | 否 | 否 |
| 差分隱私 SmartNoise 工具組 | OSS | 否 | 否 |
| 在 Azure Machine Learning 中使用自訂標籤實現資料表 | GA | 否 | 否 |
| 公平性 Azure 機器學習整合 | 公開預覽 | 否 | 否 |
| 可解釋性 SDK | GA | 是 | 是 |
| 訓練 | |||
| 測試記錄串流 | GA | 是 | 是 |
| 測試 UI | 公開預覽 | 是 | 是 |
| .NET 整合 ML.NET 1.0 | GA | 是 | 是 |
| 推斷 | |||
| 受控線上端點 | GA | 是 | 是 |
| 批次推斷 | GA | 是 | 是 |
| 其他 | |||
| 開放資料集 | 公開預覽 | 是 | 是 |
| Custom Azure AI 搜尋服務 (SDK v1) | 公開預覽 | 是 | 是 |
Azure 政府情境
| 狀況 | US-Virginia | US-Arizona | 限制 |
|---|---|---|---|
| 一般安全性設定 | |||
| 停用/控制網際網路存取 (傳入與傳出) 及特定 VNet | PARTIAL | PARTIAL | |
| 所有相關資源與服務的放置位置 | 是 | 是 | |
| 靜態與傳輸中的加密。 | 是 | 是 | |
| 對計算資源的 root 與 SSH 存取。 | 是 | 是 | |
| 維護已部署系統 (執行個體、端點等) 的安全性,包含端點防護、修補與記錄 | PARTIAL | PARTIAL | 目前不提供 VNet 後方的 ACI |
| 控制 (停用/限制) ACI/AKS 整合的使用 | PARTIAL | PARTIAL | 目前不提供 VNet 後方的 ACI |
| Azure 角色型存取控制(Azure RBAC)- 自訂角色建立 | 是 | 是 | |
| 控制 ML 服務使用的 ACR 映像存取 (Azure 提供與維護或自訂) | PARTIAL | PARTIAL | |
| 一般機器學習服務使用率 | |||
| 具備可建立模型、訓練模型、將其裝載為端點,並透過 Web 應用程式取用的開發環境 | 是 | 是 | |
| 從 ADLS(Data Lake Storage)擷取資料的能力 | 是 | 是 | |
| 從 Azure Blob 儲存體 提取數據的能力 | 是 | 是 |
其他 Azure Government 限制
對於 Azure Machine Learning 計算執行個體,在 Azure Government 中不提供可延長超過 24 小時的權杖重新整理。
模型剖析在 US-Arizona 區域不支援 4 個 CPU。
如果範例筆記本需要存取公開資料,可能無法在 Azure Government 中使用。
IP 位址:用於必要公用網際網路存取指示中的 CLI 命令不會傳回 IP 範圍。 改用 Azure IP 範圍和服務標籤來對應 Azure Government。
對於排程管線,也提供以 Blob 為基礎的觸發機制。 此機制不支援 CMK 工作區。 若要為 CMK 工作區啟用以 Blob 為基礎的觸發,需進行額外設定。 如需詳細資訊,請參閱從 Logic App 觸發機器學習管線執行 (SDK/CLI v1)。
防火牆:使用 Azure Government 區域時,請將以下主機加入防火牆設定:
- 對於 Arizona 使用:
usgovarizona.api.ml.azure.us - 對於 Virginia 使用:
usgovvirginia.api.ml.azure.us - 對於兩者:
graph.windows.net
- 對於 Arizona 使用:
Azure 由 21Vianet 營運
| 功能 | 公用雲端狀態 | CH-East-2 | CH-North-3 |
|---|---|---|---|
| 自動化機器學習 | |||
| 在 Notebooks 中建立和執行實驗 | GA | 是 | N/A |
| 在 Studio Web 體驗中建立和執行實驗 | 預覽 | 是 | N/A |
| 業界領先的預測功能 | GA | 是 | N/A |
| 深度學習和其他進階學習者的支援 | GA | 是 | N/A |
| 大型資料支援 (最高 100 GB) | 預覽 | 是 | N/A |
| Azure Databricks 整合 | GA | 是 | N/A |
| SQL、Azure Cosmos DB 和 HDInsight 整合 | GA | 是 | N/A |
| 機器學習流程 | |||
| 使用 Azure Machine Learning SDK 建立、執行並發布管線 | GA | 是 | N/A |
| 使用 Azure Machine Learning SDK 建立管道端點 | GA | 是 | N/A |
| 使用 Azure Machine Learning SDK 建立、編輯及刪除管線的排程執行 | GA | 是 | N/A |
| 在 Studio 中檢視管線執行詳細資料 | GA | 是 | N/A |
| 在 Azure Machine Learning 設計工具 中建立、執行、視覺化及發佈管道 | GA | 是 | N/A |
| Azure Databricks 與機器學習管道的整合 | GA | 是 | N/A |
| 在 Azure Machine Learning 設計工具中建立管線端點 | GA | 是 | N/A |
| 整合式筆記本 | |||
| 工作區筆記本和檔案共用 | GA | 是 | N/A |
| R 與 Python 支援 | GA | 是 | N/A |
| 虛擬網路支援 | GA | 是 | N/A |
| 計算執行個體 | |||
| 整合式筆記本的受控計算執行個體 | GA | 是 | N/A |
| Jupyter、JupyterLab 整合 | GA | 是 | N/A |
| 受控虛擬網路支援 | 預覽 | 預覽 | N/A |
| 虛擬網路 (VNet) 支援 | GA | 是 | N/A |
| SDK 支援 | |||
| Python SDK 支援 | GA | 是 | N/A |
| 安全性 | |||
| 虛擬網路(VNet)訓練支援 | GA | 是 | N/A |
| 虛擬網路(VNet)推論支援 | GA | 是 | N/A |
| 評分端點驗證 | 預覽 | 是 | N/A |
| 工作區私用端點 | GA | 是 | N/A |
| VNet 後方的 ACI | 預覽 | 否 | N/A |
| VNet 後方的 ACR | GA | 是 | N/A |
| AKS 叢集的私用 IP | 預覽 | 否 | N/A |
| 受控線上端點的網路隔離 | GA | 否 | N/A |
| 計算 | |||
| 跨工作區的配額管理 | GA | 是 | N/A |
| Kubernetes 計算 | GA | 否 | 否 |
| 機器學習的資料 | |||
| 從 SDK 建立、檢視或編輯資料集與資料存放區 | GA | 是 | N/A |
| 從 UI 建立、檢視或編輯資料集與資料存放區 | GA | 是 | N/A |
| 從 SDK 檢視、編輯或刪除資料集漂移監視器 | 預覽 | 是 | N/A |
| 從 UI 檢視、編輯或刪除資料集漂移監視器 | 預覽 | 是 | N/A |
| 機器學習生命週期 | |||
| 模型分析 | GA | 是 | N/A |
| Azure DevOps 的 機器學習 擴充功能及 Azure Machine Learning CLI | GA | 是 | N/A |
| 以 FPGA 為基礎的硬體加速模型 | 已棄用 | 已棄用 | N/A |
| Visual Studio Code 整合 | 預覽 | 否 | N/A |
| 事件方格整合 | 預覽 | 是 | N/A |
| 將 Azure 串流分析 整合於 Azure Machine Learning | 預覽 | 否 | N/A |
| 標註 | |||
| 為專案管理入口網站加標籤 | GA | 是 | N/A |
| Labeler 入口網站 | GA | 是 | N/A |
| 使用私人工作區加標籤 | GA | 是 | N/A |
| ML 輔助標註 (影像分類與物件偵測) | 預覽 | 是 | N/A |
| 負責任的 AI | |||
| UI 中的可解釋性 | 預覽 | 否 | N/A |
| 差分隱私 SmartNoise 工具組 | OSS | 否 | N/A |
| Azure Machine Learning 中的自訂標籤以實現數據表 | GA | 是 | N/A |
| 公平性 Azure Machine Learning 整合 | 預覽 | 否 | N/A |
| 可解釋性 SDK | GA | 是 | N/A |
| 訓練 | |||
| 測試記錄串流 | GA | 是 | N/A |
| 增強式學習 | 已棄用 | 已棄用 | N/A |
| 測試 UI | GA | 是 | N/A |
| .NET 整合 ML.NET 1.0 | GA | 是 | N/A |
| 推斷 | |||
| 受控線上端點 | GA | 是 | N/A |
| 批次推斷 | GA | 是 | N/A |
| Azure Stack Edge with FPGA(具 FPGA 功能) | 已棄用 | 已棄用 | N/A |
| 其他 | |||
| 開放資料集 | 預覽 | 是 | N/A |
| 自訂 Azure AI 搜尋 | 預覽 | 是 | N/A |
其他由世紀互聯提供的 Azure 限制
由 21Vianet 營運的 Azure VM SKU 有限,尤其是 GPU SKU。 僅提供 NCv3 系列 (V100)。
REST API 端點與全球 Azure 不同。 請使用以下表格尋找由 21Vianet 區域運作的 Azure REST API 端點:
REST 端點 Global Azure China-Government 管理平面 https://management.azure.com/https://management.chinacloudapi.cn/資料平面 https://{location}.experiments.azureml.nethttps://{location}.experiments.ml.azure.cnMicrosoft Entra ID https://login.microsoftonline.comhttps://login.chinacloudapi.cn若範例筆記本需要存取公用資料,可能無法運作。
IP 位址範圍:用於必要公用網際網路存取指示中的 CLI 命令不會傳回 IP 範圍。 改用 適用於由 21Vianet 操作的 Microsoft Azure 的 Azure IP 範圍和服務標籤。
目前在啟用私有端點的工作區中,Azure Machine Learning 運算實例的預覽版本不受支援,但在下一次服務部署中,CI 將在所有 Azure Machine Learning 區域得到支援。
在 Web UI 中使用中文字符搜尋成品將無法正確運作。
後續步驟
欲了解Azure Machine Learning在哪些地區可用,請參閱Products by region。