You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
|Azure AD Connect Health|Azure AD Connect Health generates alerts and reports in Azure Tables storage and blob storage.|In geo location|
89
89
|Azure AD dynamic membership for groups, Azure AD self-service group management|Azure Tables storage holds dynamic membership rule definitions.|In geo location|
90
90
|Azure AD Application Proxy|Azure AD Application Proxy stores metadata about the tenant, connector machines, and configuration data in Azure SQL.|In geo location|
91
-
|Azure AD password reset |Azure AD password reset is a back-end service using Redis Cache to track session state. To learn more, go to redis.com to see [Introduction to Redis](https://redis.io/docs/about/).|See, Intro to Redis link in center column.|
92
91
|Azure AD password writeback in Azure AD Connect|During initial configuration, Azure AD Connect generates an asymmetric keypair, using the Rivest–Shamir–Adleman (RSA) cryptosystem. It then sends the public key to the self-service password reset (SSPR) cloud service, which performs two operations: </br></br>1. Creates two Azure Service Bus relays for the Azure AD Connect on-premises service to communicate securely with the SSPR service </br> 2. Generates an Advanced Encryption Standard (AES) key, K1 </br></br> The Azure Service Bus relay locations, corresponding listener keys, and a copy of the AES key (K1) goes to Azure AD Connect in the response. Future communications between SSPR and Azure AD Connect occur over the new ServiceBus channel and are encrypted using SSL. </br> New password resets, submitted during operation, are encrypted with the RSA public key generated by the client during onboarding. The private key on the Azure AD Connect machine decrypts them, which prevents pipeline subsystems from accessing the plaintext password. </br> The AES key encrypts the message payload (encrypted passwords, more data, and metadata), which prevents malicious ServiceBus attackers from tampering with the payload, even with full access to the internal ServiceBus channel. </br> For password writeback, Azure AD Connect need keys and data: </br></br> - The AES key (K1) that encrypts the reset payload, or change requests from the SSPR service to Azure AD Connect, via the ServiceBus pipeline </br> - The private key, from the asymmetric key pair that decrypts the passwords, in reset or change request payloads </br> - The ServiceBus listener keys </br></br> The AES key (K1) and the asymmetric keypair rotate a minimum of every 180 days, a duration you can change during certain onboarding or offboarding configuration events. An example is a customer disables and re-enables password writeback, which might occur during component upgrade during service and maintenance. </br> The writeback keys and data stored in the Azure AD Connect database are encrypted by data protection application programming interfaces (DPAPI) (CALG_AES_256). The result is the master ADSync encryption key stored in the Windows Credential Vault in the context of the ADSync on-premises service account. The Windows Credential Vault supplies automatic secret re-encryption as the password for the service account changes. To reset the service account password invalidates secrets in the Windows Credential Vault for the service account. Manual changes to a new service account might invalidate the stored secrets.</br> By default, the ADSync service runs in the context of a virtual service account. The account might be customized during installation to a least-privileged domain service account, a managed service account (MSA), or a group managed service account (gMSA). While virtual and managed service accounts have automatic password rotation, customers manage password rotation for a custom provisioned domain account. As noted, to reset the password causes loss of stored secrets. |In geo location|
93
92
|Azure AD Device Registration Service |Azure AD Device Registration Service has computer and device lifecycle management in the directory, which enable scenarios such as device-state conditional access, and mobile device management.|In geo location|
94
93
|Azure AD provisioning|Azure AD provisioning creates, removes, and updates users in systems, such as software as service (SaaS) applications. It manages user creation in Azure AD and on-premises AD from cloud HR sources, like Workday. The service stores its configuration in an Azure Cosmos DB, which stores the group membership data for the user directory it keeps. Cosmos DB replicates the database to multiple datacenters in the same region as the tenant, which isolates the data, according to the Azure AD cloud solution model. Replication creates high availability and multiple reading and writing endpoints. Cosmos DB has encryption on the database information, and the encryption keys are stored in the secrets storage for Microsoft.|In geo location|
Copy file name to clipboardExpand all lines: articles/aks/cluster-autoscaler.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -30,7 +30,7 @@ Both the horizontal pod autoscaler and cluster autoscaler can also decrease the
30
30
The cluster autoscaler may be unable to scale down if pods can't move, such as in the following situations:
31
31
32
32
* A pod is directly created and isn't backed by a controller object, such as a deployment or replica set.
33
-
* A pod disruption budget (PDB) is too restrictive and doesn't allow the number of pods to be fall below a certain threshold.
33
+
* A pod disruption budget (PDB) is too restrictive and doesn't allow the number of pods to fall below a certain threshold.
34
34
* A pod uses node selectors or anti-affinity that can't be honored if scheduled on a different node.
35
35
36
36
For more information about how the cluster autoscaler may be unable to scale down, see [What types of pods can prevent the cluster autoscaler from removing a node?][autoscaler-scaledown].
@@ -19,7 +19,8 @@ Azure OpenAI provides access to many different models, grouped by family and cap
19
19
20
20
| Model family | Description |
21
21
|--|--|
22
-
|[GPT-3](#gpt-3-models)| A series of models that can understand and generate natural language. This includes the new [ChatGPT model](#chatgpt-gpt-35-turbo). |
22
+
|[GPT-4](#gpt-4-models)| A set of models that improve on GPT-3.5 and can understand as well as generate natural language and code. **These models are currently in preview.**|
23
+
|[GPT-3](#gpt-3-models)| A series of models that can understand and generate natural language. This includes the new [ChatGPT model (preview)](#chatgpt-gpt-35-turbo-preview). |
23
24
|[Codex](#codex-models)| A series of models that can understand and generate code, including translating natural language to code. |
24
25
|[Embeddings](#embeddings-models)| A set of models that can understand and use embeddings. An embedding is a special format of data representation that can be easily utilized by machine learning models and algorithms. The embedding is an information dense representation of the semantic meaning of a piece of text. Currently, we offer three families of Embeddings models for different functionalities: similarity, text search, and code search. |
25
26
@@ -53,7 +54,17 @@ You can get a list of models that are available for both inference and fine-tuni
53
54
54
55
## Finding the right model
55
56
56
-
We recommend starting with the most capable model in a model family to confirm whether the model capabilities meet your requirements. Then you can stay with that model or move to a model with lower capability and cost, optimizing around that model's capabilities.
57
+
We recommend starting with the most capable model in a model family to confirm whether the model capabilities meet your requirements. Then you can stay with that model or move to a model with lower capability and cost, optimizing around that model's capabilities.
58
+
59
+
## GPT-4 models (preview)
60
+
61
+
GPT-4 can solve difficult problems with greater accuracy than any of OpenAI's previous models. Like gpt-35-turbo, GPT-4 is optimized for chat but works well for traditional completions tasks.
62
+
63
+
These models are currently in preview. For access, existing Azure OpenAI customers can [apply by filling out this form](https://aka.ms/oai/get-gpt4).
64
+
-`gpt-4`
65
+
-`gpt-4-32k`
66
+
67
+
The `gpt-4` supports 8192 max input tokens and the `gpt-4-32k` supports up to 32,768 tokens.
57
68
58
69
## GPT-3 models
59
70
@@ -92,11 +103,11 @@ Ada is usually the fastest model and can perform tasks like parsing text, addres
The ChatGPT model (gpt-35-turbo) is a language model designed for conversational interfaces and the model behaves differently than previous GPT-3 models. Previous models were text-in and text-out, meaning they accepted a prompt string and returned a completion to append to the prompt. However, the ChatGPT model is conversation-in and message-out. The model expects a prompt string formatted in a specific chat-like transcript format, and returns a completion that represents a model-written message in the chat.
98
109
99
-
The ChatGPT model uses the same completion API that you use for other models like text-davinci-002, but it requires a unique prompt format. It's important to use the new prompt format to get the best results. Without the right prompts, the model tends to be verbose and provides less useful responses. To learn more check out our [in-depth how-to](../how-to/chatgpt.md).
110
+
To learn more about the ChatGPT model and how to interact with the Chat API check out our [in-depth how-to](../how-to/chatgpt.md).
100
111
101
112
## Codex models
102
113
@@ -173,13 +184,23 @@ When using our embeddings models, keep in mind their limitations and risks.
173
184
| davinci<sup>1</sup> | Yes | No | N/A | East US<sup>2</sup>, South Central US, West Europe<sup>2</sup> | 2,049 | Oct 2019|
174
185
| text-davinci-001 | Yes | No | South Central US, West Europe | N/A |||
175
186
| text-davinci-002 | Yes | No | East US, South Central US, West Europe | N/A | 4,097 | Jun 2021 |
176
-
| text-davinci-003 | Yes | No | East US | N/A | 4,097 | Jun 2021 |
187
+
| text-davinci-003 | Yes | No | East US, West Europe| N/A | 4,097 | Jun 2021 |
177
188
| text-davinci-fine-tune-002<sup>1</sup> | Yes | No | N/A | East US, West Europe<sup>2</sup> |||
178
-
| gpt-35-turbo<sup>3</sup> (ChatGPT) | Yes | No | N/A | East US, South Central US | 4,096 | Sep 2021
189
+
| gpt-35-turbo<sup>3</sup> (ChatGPT) (preview) | Yes | No | East US, South Central US | N/A | 4,096 | Sep 2021
179
190
180
191
<sup>1</sup> The model is available by request only. Currently we aren't accepting new requests to use the model.
181
192
<br><sup>2</sup> East US and West Europe are currently unavailable for new customers to fine-tune due to high demand. Please use US South Central region for fine-tuning.
182
-
<br><sup>3</sup> Currently, only version `"0301"` of this model is available. This version of the model will be deprecated on 8/1/2023 in favor of newer version of the gpt-35-model. See [ChatGPT model versioning](../how-to/chatgpt.md#model-versioning) for more details.
193
+
<br><sup>3</sup> Currently, only version `0301` of this model is available. This version of the model will be deprecated on 8/1/2023 in favor of newer version of the gpt-35-model. See [ChatGPT model versioning](../how-to/chatgpt.md#model-versioning) for more details.
194
+
195
+
### GPT-4 Models
196
+
197
+
| Model ID | Supports Completions | Supports Embeddings | Base model Regions | Fine-Tuning Regions | Max Request (tokens) | Training Data (up to) |
0 commit comments