Merge pull request #231806 from mrbullwinkle/mrb_03_22_2023_openai_ada_v2

[Cognitive Services] [Azure OpenAI] Update to ada_v2_model

Author: American-Dipper

articles/cognitive-services/openai/concepts/models.md

@@ -4,11 +4,11 @@ titleSuffix: Azure OpenAI
 description: Learn about the different model capabilities that are available with Azure OpenAI. 
 ms.service: cognitive-services
 ms.topic: conceptual 
-ms.date: 03/21/2023
+ms.date: 03/31/2023
 ms.custom: event-tier1-build-2022, references_regions
 manager: nitinme
-author: ChrisHMSFT
-ms.author: chrhoder
+author: mrbullwinkle #ChrisHMSFT
+ms.author: mbullwin #chrhoder
 recommendations: false
 keywords: 
 ---
@@ -128,7 +128,10 @@ Cushman is powerful, yet fast. While Davinci is stronger when it comes to analyz
 
 ## Embeddings models
 
-Currently, we offer three families of Embeddings models for different functionalities: 
+> [!IMPORTANT]
+> We strongly recommend using `text-embedding-ada-002 (Version 2)`. This model/version provides parity with OpenAI's `text-embedding-ada-002`. To learn more about the improvements offered by this model, please refer to [OpenAI's blog post](https://openai.com/blog/new-and-improved-embedding-model). Even if you are currently using Version 1 you should migrate to Version 2 to take advantage of the latest weights/updated token limit. Version 1 and Version 2 are not interchangeable, so document embedding and document search must be done using the same version of the model.
+
+Currently, we offer three families of Embeddings models for different functionalities:
 
 - [Similarity](#similarity-embedding)
 - [Text search](#text-search-embedding)
@@ -221,9 +224,13 @@ These models can only be used with Completions API requests.
 
 These models can only be used with Embedding API requests.
 
+> [!NOTE]
+> We strongly recommend using `text-embedding-ada-002 (Version 2)`. This model/version provides parity with OpenAI's `text-embedding-ada-002`. To learn more about the improvements offered by this model, please refer to [OpenAI's blog post](https://openai.com/blog/new-and-improved-embedding-model). Even if you are currently using Version 1 you should migrate to Version 2 to take advantage of the latest weights/updated token limit. Version 1 and Version 2 are not interchangeable, so document embedding and document search must be done using the same version of the model.
+
 |  Model ID  |  Base model Regions   | Fine-Tuning Regions | Max Request (tokens) | Training Data (up to)  |
 |  --- | --- | --- | --- | --- |
-| text-embedding-ada-002 | East US, South Central US, West Europe | N/A |2,046 | Sep 2021 |
+| text-embedding-ada-002 (version 2) | East US, South Central US | N/A |8,191 | Sep 2021 |
+| text-embedding-ada-002 (version 1) | East US, South Central US, West Europe | N/A |4,095 | Sep 2021 |
 | text-similarity-ada-001| East US, South Central US, West Europe | N/A | 2,046 | Aug 2020 |
 | text-similarity-babbage-001  | South Central US, West Europe | N/A | 2,046 | Aug 2020 |
 | text-similarity-curie-001 | East US, South Central US, West Europe | N/A |  2046 | Aug 2020 |

articles/cognitive-services/openai/concepts/understand-embeddings.md

@@ -7,7 +7,7 @@ manager: nitinme
 ms.service: cognitive-services
 ms.subservice: openai
 ms.topic: tutorial
-ms.date: 12/06/2022
+ms.date: 03/22/2023
 author: mrbullwinkle
 ms.author: mbullwin
 recommendations: false
@@ -28,7 +28,7 @@ Embeddings make it easier to do machine learning on large inputs representing wo
 
 One method of identifying similar documents is to count the number of common words between documents. Unfortunately, this approach doesn't scale since an expansion in document size is likely to lead to a greater number of common words detected even among completely disparate topics. For this reason, cosine similarity can offer a more effective alternative.
 
-From a mathematic perspective, cosine similarity measures the cosine of the angle between two vectors projected in a multi-dimensional space. This is beneficial because if two documents are far apart by Euclidean distance because of size, they could still have a smaller angle between them and therefore higher cosine similarity.
+From a mathematic perspective, cosine similarity measures the cosine of the angle between two vectors projected in a multi-dimensional space. This is beneficial because if two documents are far apart by Euclidean distance because of size, they could still have a smaller angle between them and therefore higher cosine similarity. For more information on cosine similarity and the [underlying formula](https://en.wikipedia.org/wiki/Cosine_similarity).
 
 Azure OpenAI embeddings rely on cosine similarity to compute similarity between documents and a query.