diff --git a/0_RAG_Overview.md b/0_RAG_Overview.md index 6e24cbc..45d3639 100644 --- a/0_RAG_Overview.md +++ b/0_RAG_Overview.md @@ -33,7 +33,7 @@ |----------------|----------------|-------------------------------| | **Retrieval** | Retrieval involves searching and extracting relevant documents or data from a knowledge base or external data source based on the input query. | Use Azure AI Search to index and query documents stored in Azure Storage Blob Containers. Configure the search index to perform semantic search and return the most relevant results. | | **Augmentation** | Augmentation involves enhancing the input query with the retrieved information to provide additional context and details. | Use Azure AI Search skillsets to preprocess the retrieved data, extracting key phrases, entities, and contextual information. This augmented input is then used to inform the generative model. | -| **Generation** | Generation involves using a generative model to process the augmented input and produce a coherent and contextually relevant response. | Deploy a generative model like GPT-4 on Azure AI Foundry. Use an Azure Function App to orchestrate the data flow, calling the Azure AI Foundry model API to generate responses based on the augmented input. | +| **Generation** | Generation involves using a generative model to process the augmented input and produce a coherent and contextually relevant response. | Deploy a generative model like GPT-6 Astra on Azure AI Foundry. Use an Azure Function App to orchestrate the data flow, calling the Azure AI Foundry model API to generate responses based on the augmented input. | > Implementing RAG Pattern with Azure AI: @@ -44,7 +44,7 @@ graph LR 1. **Set Up a Knowledge Base**: Store your documents in Azure Storage Blob Containers or another accessible data source. 2. **Configure a Retrieval System**: Use Azure AI Search to index and retrieve relevant documents based on user queries. -3. **Integrate with a Generative Model**: Use a generative model like GPT-4 to process the retrieved documents and generate responses. +3. **Integrate with a Generative Model**: Use a generative model like GPT-6 Astra to process the retrieved documents and generate responses. > Traditional methods, `Retrieval-Augmented Generation (RAG)`, and `Agentic RAG`: @@ -74,7 +74,7 @@ graph LR E --> F[Contextual Info] E --> G[Enhanced Query] A --> H[Generation] - H --> I[LLM: e.g GPT-4] + H --> I[LLM: e.g. GPT-6 Astra] H --> J[Coherent Response] A --> K[Applications] K --> L[Question Answering] diff --git a/GPT-RAG_SolutionAccelerator/docs/AUTOMATED_INSTALLATION.md b/GPT-RAG_SolutionAccelerator/docs/AUTOMATED_INSTALLATION.md index b60da78..6a82111 100644 --- a/GPT-RAG_SolutionAccelerator/docs/AUTOMATED_INSTALLATION.md +++ b/GPT-RAG_SolutionAccelerator/docs/AUTOMATED_INSTALLATION.md @@ -88,7 +88,7 @@ Here is the complete list of resources for a standard Zero Trust deployment, inc
Generates responses and vector embeddings. - SKU: Standard - Deployments: - - Regional gpt-4o, 40 TPM. + - Regional gpt-6-astra, 40 TPM. - text-embedding-3-large, 40 TPM. - **Search Service**
Provides vector indexes for the retrieval step. diff --git a/GPT-RAG_SolutionAccelerator/docs/CUSTOMIZATIONS_BYOR.md b/GPT-RAG_SolutionAccelerator/docs/CUSTOMIZATIONS_BYOR.md index db90d76..bf3820d 100644 --- a/GPT-RAG_SolutionAccelerator/docs/CUSTOMIZATIONS_BYOR.md +++ b/GPT-RAG_SolutionAccelerator/docs/CUSTOMIZATIONS_BYOR.md @@ -103,7 +103,7 @@ If you're using a different name for the Embedding model, you'll need to set the | Embeddings Deployment Name | AZURE_EMBEDDINGS_DEPLOYMENT_NAME | | Embeddings Vector Size | AZURE_EMBEDDINGS_VECTOR_SIZE | -The default GPT model is **gpt-4o** version **2024-11-20** and the default deployment name is **chat**. If you're using the pre-created service with these default settings, no further modifications are required. +The default GPT model is **gpt-6-astra** and the default deployment name is **chat**. If you're using the pre-created service with these default settings, no further modifications are required. However, if you're using a different name for the GPT, or a different model, you'll need to set the corresponding environment variables as shown in the table below. diff --git a/GPT-RAG_SolutionAccelerator/docs/GUIDE.md b/GPT-RAG_SolutionAccelerator/docs/GUIDE.md index ed68556..dff24e3 100644 --- a/GPT-RAG_SolutionAccelerator/docs/GUIDE.md +++ b/GPT-RAG_SolutionAccelerator/docs/GUIDE.md @@ -344,7 +344,7 @@ As of December 19, 2024, the solution has been tested in the following regions w #### Azure OpenAI Model Support -By default, the solution deploys a **Global Deployment** of the GPT-4o model (version 2024-11-20) and a **Standard Deployment** of the text-embedding-3-large model for embedding generation. To ensure compatibility, refer to the Azure OpenAI model summary table and region availability at the following link: +By default, the solution deploys a **Global Deployment** of the GPT-6 Astra model and a **Standard Deployment** of the text-embedding-3-large model for embedding generation. To ensure compatibility, refer to the Azure OpenAI model summary table and region availability at the following link: [Azure OpenAI Model Availability](https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models?tabs=global-standard%2Cstandard-chat-completions#model-summary-table-and-region-availability). If your selected region does not support these default models, you have two options: @@ -355,7 +355,7 @@ Before proceeding with the [Provision Infrastructure Components](#9-provision-in For the Chat Completions model: ```bash - azd env set AZURE_CHAT_GPT_MODEL_NAME gpt-4o + azd env set AZURE_CHAT_GPT_MODEL_NAME gpt-6-astra azd env set AZURE_CHAT_GPT_MODEL_VERSION 2024-11-20 azd env set AZURE_CHAT_GPT_DEPLOYMENT_TYPE GlobalStandard ``` @@ -1045,7 +1045,7 @@ To learn how this scenario works and configure it, check the [NL2SQL and Fabric ## Enabling Multimodality -To enable GPT-RAG to use multimodal capabilities, such as those provided by GPT-4o, set the `MULTIMODAL` environment variable to `true` in the data ingestion and set the `multimodal_rag` agent strategy in agentic orchestration Function Apps. For more details on how multimodality works and image data is ingested, refer to [Multimodal RAG Overview](MULTIMODAL_RAG.md) and the documentation in the data ingestion repository: [Multimodal Ingestion](https://github.com/Azure/gpt-rag-ingestion?tab=readme-ov-file#multimodal-ingestion) respectively. +To enable GPT-RAG to use multimodal capabilities, such as those provided by GPT-6 Astra, set the `MULTIMODAL` environment variable to `true` in the data ingestion and set the `multimodal_rag` agent strategy in agentic orchestration Function Apps. For more details on how multimodality works and image data is ingested, refer to [Multimodal RAG Overview](MULTIMODAL_RAG.md) and the documentation in the data ingestion repository: [Multimodal Ingestion](https://github.com/Azure/gpt-rag-ingestion?tab=readme-ov-file#multimodal-ingestion) respectively. > [!NOTE] > Currently, only the Agentic Orchestrator supports this feature. @@ -1219,7 +1219,7 @@ Here is the complete list of resources for a standard Zero Trust deployment, inc
Generates responses and vector embeddings. - SKU: Standard - Deployments: - - Regional gpt-4o, 40 TPM. + - Regional gpt-6-astra, 40 TPM. - text-embedding-3-large, 40 TPM. - **Search Service**
Provides vector indexes for the retrieval step. diff --git a/GPT-RAG_SolutionAccelerator/docs/MANUAL_ENVIRONMENT.md b/GPT-RAG_SolutionAccelerator/docs/MANUAL_ENVIRONMENT.md index 1d8d255..652910c 100644 --- a/GPT-RAG_SolutionAccelerator/docs/MANUAL_ENVIRONMENT.md +++ b/GPT-RAG_SolutionAccelerator/docs/MANUAL_ENVIRONMENT.md @@ -23,7 +23,7 @@ az functionapp config appsettings set --name --resource-grou az functionapp config appsettings set --name --resource-group --settings AZURE_OPENAI_CHATGPT_LLM_MONITORING="true" -az functionapp config appsettings set --name --resource-group --settings AZURE_OPENAI_CHATGPT_MODEL="gpt-4o" +az functionapp config appsettings set --name --resource-group --settings AZURE_OPENAI_CHATGPT_MODEL="gpt-6-astra" az functionapp config appsettings set --name --resource-group --settings AZURE_OPENAI_EMBEDDING_DEPLOYMENT="text-embedding-3-large" diff --git a/GPT-RAG_SolutionAccelerator/docs/MANUAL_INSTALLATION.md b/GPT-RAG_SolutionAccelerator/docs/MANUAL_INSTALLATION.md index c5957e4..a43c073 100644 --- a/GPT-RAG_SolutionAccelerator/docs/MANUAL_INSTALLATION.md +++ b/GPT-RAG_SolutionAccelerator/docs/MANUAL_INSTALLATION.md @@ -66,7 +66,7 @@ Here is the complete list of resources for a standard Zero Trust deployment, inc
Generates responses and vector embeddings. - SKU: Standard - Deployments: - - Regional gpt-4o, 40 TPM. + - Regional gpt-6-astra, 40 TPM. - text-embedding-3-large, 40 TPM. - **Search Service**
Provides vector indexes for the retrieval step. @@ -227,7 +227,7 @@ Gather Necessary Information: - **Azure OpenAI** - Create an Azure OpenAI service - Create deployments: - - Regional gpt-4o, 40 TPM. + - Regional gpt-6-astra, 40 TPM. - text-embedding-3-large, 40 TPM. - Disable public network access. diff --git a/GPT-RAG_SolutionAccelerator/docs/MULTIMODAL_RAG.md b/GPT-RAG_SolutionAccelerator/docs/MULTIMODAL_RAG.md index 33aa81c..eeb9d07 100644 --- a/GPT-RAG_SolutionAccelerator/docs/MULTIMODAL_RAG.md +++ b/GPT-RAG_SolutionAccelerator/docs/MULTIMODAL_RAG.md @@ -41,9 +41,9 @@ This document outlines the architecture and workflow for our **multimodal Retrie - Perform retrieval, searching both `contentVector` and `captionVector` fields. - Results include both textual context and references to relevant images. -6. **Response Generation (GPT-4o)** +6. **Response Generation (GPT-6 Astra)** - Build a multimodal prompt that includes retrieved text, image descriptions, and image URLs. - - GPT-4 generates a final enriched response, referencing both textual and visual elements. + - GPT-6 Astra generates a final enriched response, referencing both textual and visual elements. 7. **Document & Image Lifecycle Management** - Deleting a document from the index triggers an Azure Function to remove associated images from Blob Storage. diff --git a/GPT-RAG_SolutionAccelerator/docs/RAG_CONCEPTS.md b/GPT-RAG_SolutionAccelerator/docs/RAG_CONCEPTS.md index bbd3596..276b15c 100644 --- a/GPT-RAG_SolutionAccelerator/docs/RAG_CONCEPTS.md +++ b/GPT-RAG_SolutionAccelerator/docs/RAG_CONCEPTS.md @@ -16,7 +16,7 @@ Reference implementation of the Retrieval-Augmented Generation (RAG) pattern. * [Retrieval Augmented Generation (RAG) in Azure AI Search](https://learn.microsoft.com/en-us/azure/search/retrieval-augmented-generation-overview) -* [Build and maintain your company Copilot with Azure ML and GPT-4](https://www.youtube.com/watch?si=B2tjsq4z4R7rksEw&v=2meEvuWAyXs) +* [GPT-6 Astra in Microsoft Foundry](https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure) * [Revolutionize your Enterprise Data with ChatGPT: Next-gen Apps w/ Azure OpenAI and AI Search](https://aka.ms/entgptsearchblog) diff --git a/README.md b/README.md index 0162606..41cca92 100644 --- a/README.md +++ b/README.md @@ -22,8 +22,7 @@ - [Azure Developer CLI (azd)](https://learn.microsoft.com/en-us/azure/developer/azure-developer-cli/) - [What is PowerShell?](https://learn.microsoft.com/en-us/powershell/scripting/overview?view=powershell-7.5) - [Integrate Azure services with virtual networks for network isolation](https://learn.microsoft.com/en-us/azure/virtual-network/vnet-integration-for-azure-services) -- [GPT-4 and GPT-4 Turbo models max tokens](https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models?tabs=global-standard%2Cstandard-chat-completions#gpt-4-and-gpt-4-turbo-models) -- [GPT-4 and GPT-4 Turbo model availability](https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models?tabs=standard%2Cstandard-chat-completions#gpt-4-and-gpt-4-turbo-model-availability) +- [GPT-6 Astra model limits and availability](https://learn.microsoft.com/en-us/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure) - [View the service principal for a managed identity using the Azure portal](https://learn.microsoft.com/en-us/entra/identity/managed-identities-azure-resources/how-to-view-managed-identity-service-principal?pivots=identity-mi-service-principal-portal#view-the-service-principal-for-a-managed-identity-using-the-azure-portal) - [Microsoft.Search searchServices bicep template](https://learn.microsoft.com/en-us/azure/templates/microsoft.search/searchservices?pivots=deployment-language-bicep) - [Recover/Purge a deleted resource](https://learn.microsoft.com/en-us/azure/ai-services/recover-purge-resources?tabs=azure-portal#recover-a-deleted-resource)