Category: Microsoft
Category Archives: Microsoft
third party Extensions are getting turned off by edge
i have installed extensions which is only available in chrome but not in edge and made it not uninstallable using registry edit using policy “ExtensionInstallForcelist” but when i am closing the edge with task manage by end task. and again turning on the edge i see the extensions get turned off even i have allowed the “Allow extensions from other stores” please fix this issue due to this i am able to bypass the adult blocker extension many people also reported this https://drive.google.com/file/d/1I8-coZr9WnIoovO4xiF7BxICMoYYMSU4/view?usp=drivesdk
Edge keeps turning off my Chrome extensions – Microsoft Community also many people reported this issue in forum you can see
i have installed extensions which is only available in chrome but not in edge and made it not uninstallable using registry edit using policy “ExtensionInstallForcelist” but when i am closing the edge with task manage by end task. and again turning on the edge i see the extensions get turned off even i have allowed the “Allow extensions from other stores” please fix this issue due to this i am able to bypass the adult blocker extension many people also reported this https://drive.google.com/file/d/1I8-coZr9WnIoovO4xiF7BxICMoYYMSU4/view?usp=drivesdk Edge keeps turning off my Chrome extensions – Microsoft Community also many people reported this issue in forum you can see Read More
More Performance + AI Integration | Azure Database for MySQL - Flexible Server
Bring your MySQL workloads to run on Azure. Azure Database for MySQL — Flexible Server offers a powerful, fully managed solution for MySQL workloads, providing unique platform-level optimizations that significantly enhance connection scaling and cost performance. Ensure high efficiency and reduced latency for mission-critical applications for up to twice the performance of other offerings. The integration with Azure OpenAI Service and Azure AI Search further extends its capabilities, enabling intelligent vector-based search and generative AI responses for more accurate and relevant user queries.
Join Parikshit Savjani, Azure Database for MySQL Principal Group PM, shares how Azure Database for MySQL — Flexible Server transforms traditional MySQL applications into high-performance, intelligent systems by combining the scalability and security of Azure with advanced AI-driven insights.
Run MySQL on Azure.
A fully managed service with unique optimizations — doubling performance, and enabling AI-driven search and responses. See it here.
Redirect transaction logging to high-speed SSDs.
Higher throughput, reduced latency, & higher cost efficiency during traffic spikes. Check it out.
Enable semantic search in Azure Database for MySQL.
Relevant and intelligent search results with Azure AI Search and OpenAI services. Take a look.
Watch our video here:
QUICK LINKS:
00:00 — Run MySQL on Azure
00:56 — Run your LAMP stack on Azure
01:24 — Accelerated Logs capability
03:02 — Vector-based search and generative AI
04:02 — Leverage Azure AI Search & Azure OpenAI services
04:29 — Establish database connection
05:52 — Build vector index
07:04 — Create an Indexer
07:32 — Semantic Search
07:54 — Test it in the Azure AI playground
08:49 — Wrap up
Link References
Check out https://aka.ms/mysql-resources
Join the community for Azure Database for MySQL at https://aka.ms/mysql-contributors
Unfamiliar with Microsoft Mechanics?
As Microsoft’s official video series for IT, you can watch and share valuable content and demos of current and upcoming tech from the people who build it at Microsoft.
Subscribe to our YouTube: https://www.youtube.com/c/MicrosoftMechanicsSeries
Talk with other IT Pros, join us on the Microsoft Tech Community: https://techcommunity.microsoft.com/t5/microsoft-mechanics-blog/bg-p/MicrosoftMechanicsBlog
Watch or listen from anywhere, subscribe to our podcast: https://microsoftmechanics.libsyn.com/podcast
Keep getting this insider knowledge, join us on social:
Follow us on Twitter: https://twitter.com/MSFTMechanics
Share knowledge on LinkedIn: https://www.linkedin.com/company/microsoft-mechanics/
Enjoy us on Instagram: https://www.instagram.com/msftmechanics/
Loosen up with us on TikTok: https://www.tiktok.com/@msftmechanics
Video Transcript:
-If MySQL is your app data tier of choice, you will want to run it on Azure. In the next few minutes, I’ll show you how Azure Database for MySQL Flexible Server, as a fully-managed service, builds on top of the community-driven MySQL to provide unique platform-level optimizations where, for example, using the new Accelerated Logs capability, you can significantly improve connection scaling and cost performance for your web apps, with up to twice performance compared to other offerings.
-Or you can combine Azure Database for MySQL with Azure AI Search and Azure Open AI Service to light up intelligent vector search for natural language querying and generative AI responses, with your existing web and e-commerce apps, all while operating on an enterprise-grade foundation of advanced platform-level security and resiliency across Azure’s global network backbone, and taking advantage of built-in AI-powered technical guidance with Copilot in Azure. Microsoft, in fact, is a major open-source contributor, and as an open platform, you can run your entire LAMP stack on Azure.
-This is literally pure MySQL on Azure, with community updates implemented within months of being available for a high level of compatibility and extension support, with a few added advantages, where we give you platform-level optimizations for your workloads running on Azure.
-For example, here I’ve deployed the LAMP base Magento framework for my e-commerce app, and with our new Accelerated Logs capability in the Business Critical Service Tier of Azure Database for MySQL Flexible Server, I’m going to show you how we built on top of MySQL’s robust scale-out architecture to improve cost performance with reduced latency for disk writes.
-This e-commerce app is prone to bursts of traffic whenever there is a specific promotion or event, and so I’m going to use the sysbench tool for around 300 seconds, 128 threads, 20 tables, and 20 million records per table to simulate high throughput traffic to the database and measure the number of transactions and queries per second.
-I’ll first run this test without Accelerated Log enabled, and you’ll see that measured throughput is around 4,500 transactions per second, 90,000 queries per second, and latency is around 51 milliseconds at the 95th percentile.
-Let’s now go back to the Azure portal to enable Accelerated Logs. And in sysbench, we’ll run the exact same test again and let the transactions process.
-Once it’s completed, you’ll see that we have doubled the throughput with 9,500 transactions per second versus 4,500 before, and 190,000 queries per second versus 90,000 before. And the response time is reduced by more than half with latency measured at 22 milliseconds at the 95th percentile versus 51 milliseconds before.
-We are able to achieve this through Azure’s platform-level optimizations, where under the covers, we keep regular data file IO operations where they are, but redirect IO-intensive transaction logging to higher speed SSDs for lower latency and higher performance.
-This ensures better responsiveness, higher throughput, and cost efficiency for your mission-critical workloads. Next, let’s make the same workload more intelligent with vector-based search and generative AI responses. This is the front end of my e-commerce app. Its search is keyword-based and that has limitations. You can see here I’m searching for a rain jacket for women, and because it’s searching on jacket as a keyword, not only is it returning the result for women’s jacket, but also for men, and not all jackets are necessarily rain jackets.
-Now, let me show you the same thing using semantic search with the responses returned using a custom generative AI experience. I’ll use natural language to type, “We live in Seattle and I’m looking for a rain jacket for my wife which is highly rated and recommended by other customers.”
-As you can see, the Magento Product Recommender Chat was able to interpret my requirements in natural language, perform semantic search from the product and reviews data stored in our backend MySQL database, and summarize the review results back to me to recommend the Inez Full-Zip Jacket.
-Let’s see what is happening on the backend. To support semantic search for our e-commerce app, we leverage Azure AI Search and Azure OpenAI services. Azure AI Search pulls the product and reviews data from the backend MySQL database for Magento, using an indexer that runs periodically. The reviews data is further chunked and vectorized using Azure OpenAI’s text embedding model. In Azure AI Search, the vectorized data is then persisted in a vector search index. In fact, let me show you how we implement this architecture in code.
-Here we use Python code in Jupyter Notebook to define, build, and refresh vector search index in Azure AI Search. And the first thing we need to do is establish our database connection. To do that, I’ve defined the connection strength for Azure Database for MySQL Flexible Server hosting Luma’s Magento e-commerce web database.
-Next, we need to connect to our Azure AI services that we’ll use to generate embeddings. Here I’ve entered the Azure OpenAI deployment details, including the API base, API key, and API type, which in this case is Azure. I’m using OpenAI’s text embedding ada-002 as the embedding model, with embedding size of 1,536 dimension, and the GPT-4 large language model. Now I have the connections in place to start building a vector index.
-I’ll use the service endpoint for Azure AI Search. Here I’ve defined the search key, the service endpoint details, and my index name that I’ll use later. Now, I need to define the query that the AI search will use to populate the vector index. In this case, because the products and customer reviews data are stored in separate tables in the Magento MySQL database, I’ve defined a single-view abstraction name, product_review_data_all in the database. This joins the product and reviews table to pull all the products information with their respective reviews.
-Now, I can complete the configuration of the vector index that I named earlier. Because customer review text strings can be quite long and span multiple themes, we need to break those into smaller pieces or chunks so that each main theme in any given review gets its own unique vector embedding. I’ve done that here, and each chunk will get a new row. This is represented by a unique review ID field. Below that, I’ve added seven additional parent fields to each new row to provide more context. We’ll define the process that splits the review text into chunks in a moment.
-For the vector index itself, I’ve used Hierarchical Navigable Small World, or HNSW graphs, which is one of the most popular algorithms for similarity search. I’ve defined the HNSW parameters here, and I’ve defined the connection settings for the VectorSearchProfile and vectorizers that I’ll use in a moment.
-Now, for the vector index, I’ve defined the search types. Here I’m using vector search as well as semantic search, which together produce what’s called hybrid search. It means you can search using keywords or natural language descriptions.
-So I’ll go ahead and run it. Once it completes, you’ll see our Magento review index is now created. Next, we create an indexer that will pull the product reviews data from our MySQL database, split those into chunks, generate the vectors for each chunk, and write those to the search index.
-Here we are using two pre-built skillsets. the Split Skill generates chunks from each review. Azure OpenAI’s Embedding Skill then uses those chunks to generate the vector embedding values for the index. Now we can run everything to insert the vector data into our vector index. Our Azure AI Search service is now ready for semantic search queries. In fact, I can try this out from the notebook.
-I’ll execute a semantic search query, “suggest me some rain jackets for women,” against our vector search index, and you’ll see the same results that we saw before in our web app. The Inez Full-Zip Jacket comes with the highest semantic search score and is our top result. And with our index running, we can also try this out in the Azure AI studio playground. To do that, I’ll just need to connect it to my search index. From setup, in the Add Your Data tab, I’ll add my data source. In this case, it’s the Azure AI Search type we just created. Then I need to select my index. I’ll enable the vector search option and then use the ada-002 embedding model.
-I’ll keep the search type as hybrid plus semantic, and that’s it. From here, we are ready to start configuring the GenAI experience in the chat playground. Using the system message, I can instruct the OpenAI GPT-4 model to behave as a product recommender assistant. Now, to test it out, I’ll paste in the same query from before and run it. Then as it completes, you’ll see the Inez Full-Zip Jacket shows up as the top recommended rain jacket, along with a brief summary of the reviews. Now with everything tested, we are ready to run it in production, like you saw on our website earlier.
-So that was an overview of the unique advantages of bringing your MySQL workloads on Azure, from its performance optimizations, to how you can extend it with Azure AI services. To learn more, please check out aka.ms/mysql-resources. And you can join the community for Azure Database for MySQL at aka.ms/mysql-contributors. And keep watching Microsoft Mechanics for the latest tech updates. Thanks for watching.
Microsoft Tech Community – Latest Blogs –Read More
Introducing new troubleshooting guides features
We are excited to announce the release of new features aimed at enhancing the troubleshooting experience in Azure PostgreSQL Flexible Server. These features provide detailed insights and metrics, enabling you to optimize your database performance effectively.
Here are the use cases where these features can be used:
Enhancements to autovacuum monitoring:
One of the significant enhancements is the addition of table-level autovacuum details. This feature allows you to:
Monitor the number of times autovacuum ran.
Check the number of dead tuples cleaned or not removed.
Track the time taken to run the autovacuum.
Furthermore, we now display the configuration of some important parameters which control autovacuum. We validate some of those parameters and warn you if we consider that the existing configuration might cause issues with the correct functioning of autovacuum. We have also added a new Enhanced metrics tab which displays the Maximum number of Transaction IDs in use and the Oldest backend XMin metrics to provide a comprehensive view of autovacuum health.
Enhancements to the CPU troubleshooting guide:
To help you manage high CPU usage, we have introduced several new features:
Parameter check for excessive logging that might impact the CPU usage and overall server performance.
New “Top queries by total duration” tab which displays Query Store queries with total time.
New insight card showing sessions whose state is idle and have an open transaction.
New tab to showcase PgBouncer metrics and best practices:
A new tab to display the locking information: In this section, you can get the overview of locks in your server.
You can get the count of sessions involved in Locks.
You can monitor the acquired locks duration during the specified time.
You can also get insights on blocking session where sessions are still waiting for lock acquisition.
Enhancements to the memory troubleshooting guide:
For troubleshooting high memory usage, we have added validation for memory parameters where it can show you an insight when you change these parameters higher than defaults.
We have also introduced an insight for long (>24hrs) idle sessions that might consume server resources.
We believe these new features will significantly enhance your ability to troubleshoot and optimize your Azure PostgreSQL Flexible Server. We look forward to your feedback!
Getting started:
To learn more about the troubleshooting guides, visit our documentation below:
What are the troubleshooting guides in Azure Database for PostgreSQL Flexible Server
How to use the troubleshooting guides for Azure Database for PostgreSQL Flexible Server
Microsoft Tech Community – Latest Blogs –Read More
List audit report
Hello everyone,
I have a Sharepoint list that is updated by a group of users, I wonder if there is a way to know who was the last user to update it.
Is that possible?
Thank you,
Hello everyone, I have a Sharepoint list that is updated by a group of users, I wonder if there is a way to know who was the last user to update it. Is that possible? Thank you, Read More
Indexing option always index’s my clone of my C drive which is F, can’t index C: drive
I really don’t want to open the case to unplug the drive.
How do I make it index my main c: drive?
When I select for C:, indexing options select clone of my OS drive to index.
No wonder can not find anything in file manager searches
When I unselect F: and select C:, it reselects F:
I really don’t want to open the case to unplug the drive.How do I make it index my main c: drive?When I select for C:, indexing options select clone of my OS drive to index.No wonder can not find anything in file manager searchesWhen I unselect F: and select C:, it reselects F: Read More
Syksyn Kumppanitunnit 2024
SYKSYN 2024 KUMPPANITUNNIT
Huomaattehan, että jo pidettyjen Kumppanituntien tallenteet löytyvät alta samasta rekisteröitymislinkistä noin kahden tunnin kuluttua Kumppanitunnin jälkeen. Kevään 2024 Kumppanituntien tallenteet löytyvät tästä linkistä.
Microsoftin tekniset Kumppanitunnit järjestetään Cloud Champion -sivustolla, josta ne ovat kätevästi saatavilla tallenteina pari tuntia live-lähetyksen jälkeen. Muistathan rekisteröityä Cloud Champion -alustalle ensimmäisellä kerralla, jonka jälkeen pääset aina sisältöön sekä tallenteisiin käsiksi. Pääset rekisteröitymään, “Register now”-kohdasta. Täytä tietosi ja valitse Distributor kohtaan – Other, mikäli et tiedät tukkurianne.
Kumppanitunti on joka toinen perjantai klo 10.00-11.00 järjestettävä Microsoftin tekninen kumppaniwebinaari, joka on tarkoitettu Microsoftin kumppaneille. Webinaarissa keskitymme Microsoftin ratkaisualueiden teknologioiden mielenkiintoisiin uutuuksiin. Microsoftin suomalaiset arkkitehdit, tuotepäälliköt ja teknologiastrategit ovat poimineet kiinnostavia ja hyödyllisiä aiheita, joita he vuorollaan esittelevät. Lisäksi luodaan alkuun lyhyt katsaus MAICPP-kumppaniohjelman uutisiin sekä tärkeisiin koulutuksiin. Webinaarissa on mahdollisuus esittää myös kysymyksiä aiheesta. Tee itsellesi toistuva kalenterivaraus ja muista ilmoittautua alla olevista linkeistä.
Alla on alustavat teemat syksylle 2024, joiden tarkka agenda päivitetään lähempänä esityspäivää, sekä lisätään linkit aiempien Kumppanituntien tallenteisiin.
16.8.2024 Microsoft AI- ja Copilot-strategian päivitys kumppaneille
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja: Vesa-Matti Paananen, teknologiajohtaja, Microsoft
30.8.2024 Tekoäly tukemassa ERP-ratkaisuja
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja: Tiia Kuokkanen, ratkaisuarkkitehti, Microsoft
13.9.2024 Tietoturva osana AI-governanssia
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja:
27.9.2024 Data-alustojen migraatio Microsoft Fabricilla
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja:
11.10.2024 Azure Open AI -ratkaisujen joustava kehitys ja kapasiteetin turvaaminen
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja:
25.10.2024 Tietoturva osana pilvimigraatiota
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja:
8.11.2024 Tietosuojapäivitys, vastuullinen tekoäly ja EU AI ACT
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja:
22.11.2024 Poimintoja Ignitestä – Kumppanitiimin tärpit
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja:
5.12.2024 Copilot for Microsoft 365 uudet ominaisuudet
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Huom! Torstai itsenäisyys päivän johdosta
Puhuja:
20.12.2024 Kumppanitunnin jouluspesiaali
Rekisteröidy tästä linkistä (linkki tulee lähempänä päivää).
Puhuja:
Löydät kevään 2024 Kumppanituntien tallenteet ja materiaalit tästä.
Microsoft Tech Community – Latest Blogs –Read More
Recording – Lowering Overhead Burden and Increasing Productivity of Medical Affairs Directors/Medica
“Recording – Lowering Overhead Burden and Increasing Productivity of Medical Affairs Directors/Medical Science Liaisons – Starter Series Round 2 Pharma Focus”
Copilot for Microsoft 365 is a digital assistant that can help Medical Affairs Directors and Medical Science Liaisons manage their work such as:
Streamline Presentations
Better prepare for KOL Engagements
Easily stay current with the latest publications and clinical trial data related to the disease area you support
Staying on top of team meetings
ISS Assessments
Clinical Trial Calls
In this webinar recording Microsoft’s Helen Guan and Michael Gannotti present how Copilot for Microsoft 365 can benefit medical affairs directors and medical science liaisons. Additionally, Helen takes viewers through two separate use cases with demos of addressing these professional needs.
To See the Entire Webinar Series click here.
Resources:
Presentation slide deck in PDF format
HLS Copilot Snacks (microsoft.com)
Copilot for Microsoft 365 – Microsoft Adoption
Copilot Success Kit – Microsoft Adoption
Copilot in Word help & learning (microsoft.com)
Copilot in PowerPoint help & learning (microsoft.com)
Copilot in OneNote help & learning (microsoft.com)
Copilot in Outlook help & learning (microsoft.com)
Copilot in Excel help & learning (microsoft.com)
Copilot in Microsoft Teams help & learning
Microsoft Copilot grounded in your work data
Copilot in Microsoft Loop help & learning
Welcome to Copilot in Windows – Microsoft Support
Copilot in Whiteboard help & learning (microsoft.com)
Welcome to Copilot in Forms – Microsoft Support
Learn about Copilot prompts – Microsoft Support
Copilot prompts Toolkit
Data, Privacy, and Security for Microsoft Copilot for Microsoft 365 | Microsoft Learn
To See the Entire Webinar Series click here.
Thanks for visiting – Michael Gannotti LinkedIn
Microsoft Tech Community – Latest Blogs –Read More
Creating Logic App to Identify Low Storage Devices from Intune
Hello everyone,
I’m seeking some assistance with creating a Logic App. I need to identify devices in Intune that have 5GB or less of available space and receive an email with the details of these devices, including their names.
Is this achievable?
Hello everyone, I’m seeking some assistance with creating a Logic App. I need to identify devices in Intune that have 5GB or less of available space and receive an email with the details of these devices, including their names.Is this achievable? Read More
Windows Defender Advanced Threat Protection Service failed to start
Good Day,
I have one PC Microsoft Windows 11 Enterprise couldn’t be onboarded because Windows Defender Advanced Threat Protection Network Detection and Response could not be started. Event ID 101 is being logged under Microsoft-Windows-SENSE/Operational with below error “Windows Defender Advanced Threat Protection Network Detection and Response executable failed to start. Failure code: 0x8007051A”
Thanks
Good Day,I have one PC Microsoft Windows 11 Enterprise couldn’t be onboarded because Windows Defender Advanced Threat Protection Network Detection and Response could not be started. Event ID 101 is being logged under Microsoft-Windows-SENSE/Operational with below error “Windows Defender Advanced Threat Protection Network Detection and Response executable failed to start. Failure code: 0x8007051A” Thanks Read More
Outlook Prints Current Date Instead of Start and End Date for the MS Purview Calendar Type Files
Hi Team,
We are facing the below issue with MS Purview Calendar Type Files While Opening it in the Outlook Application, Outlook Prints the Current Date Instead of the Start and End Date for the MS Purview Calendar Type Files all other normal Calendar File Show the Date Proper.
You can find more details on it using the below link that we have already generated in the Microsoft Community.
Ticket Generated in the Microsoft Community
Please help me out with this, and suggest if there is any solution to resolve it.
Hi Team,We are facing the below issue with MS Purview Calendar Type Files While Opening it in the Outlook Application, Outlook Prints the Current Date Instead of the Start and End Date for the MS Purview Calendar Type Files all other normal Calendar File Show the Date Proper. You can find more details on it using the below link that we have already generated in the Microsoft Community.Ticket Generated in the Microsoft Community Please help me out with this, and suggest if there is any solution to resolve it. Read More
Unable to add account in outlook in Mac computer
Hello
Please i need your help on this issue.
One of my customer is unable to add account in outlook in Mac computer error says account exist already. If he signs into Windows system with the same account it shows the account is there i.e. we can see the account.
If he signs into outlook web we can still see that the email account is there meaning it has actually been added
But does not show up in outlook in the Mac computer.
Hello Please i need your help on this issue. One of my customer is unable to add account in outlook in Mac computer error says account exist already. If he signs into Windows system with the same account it shows the account is there i.e. we can see the account.If he signs into outlook web we can still see that the email account is there meaning it has actually been addedBut does not show up in outlook in the Mac computer. Read More
Can M365 Copilot Business Users Still Use GPT Builder for Custom GPTs?
Hi everyone,
I’ve come across some conflicting information and I need some clarity. I’m a Microsoft 365 Copilot Business user and I’m trying to figure out if I can still use the GPT Builder to create custom GPTs.
From what I understand, only the consumer version, “Copilot Pro,” has had the GPT Builder functionality shut down. Here’s what I’ve gathered:
Copilot Pro: This subscription is tied to a simple Microsoft account and is primarily for private users. Microsoft has indeed disabled the GPT Builder for these users.Copilot for Microsoft 365: This can be added to a Microsoft 365 Business account on a per-user basis. From my understanding, business customers can still utilize the GPT Builder.
Currently, when I try to create a custom GPT via the web interface as a business user, I no longer see the option available, similar to the experience of private consumers.
Could anyone confirm if as a Microsoft 365 Copilot Business user, I still have access to the GPT Builder for creating custom GPTs? Any official documentation or personal experiences would be greatly appreciated!
Thanks in advance!
Hi everyone,I’ve come across some conflicting information and I need some clarity. I’m a Microsoft 365 Copilot Business user and I’m trying to figure out if I can still use the GPT Builder to create custom GPTs.From what I understand, only the consumer version, “Copilot Pro,” has had the GPT Builder functionality shut down. Here’s what I’ve gathered:Copilot Pro: This subscription is tied to a simple Microsoft account and is primarily for private users. Microsoft has indeed disabled the GPT Builder for these users.Copilot for Microsoft 365: This can be added to a Microsoft 365 Business account on a per-user basis. From my understanding, business customers can still utilize the GPT Builder. Currently, when I try to create a custom GPT via the web interface as a business user, I no longer see the option available, similar to the experience of private consumers.Could anyone confirm if as a Microsoft 365 Copilot Business user, I still have access to the GPT Builder for creating custom GPTs? Any official documentation or personal experiences would be greatly appreciated! Thanks in advance! Read More
No option to remove personal email from the computer
Hello
Please i need your help on this issue.
There is no option to remove personal email from the computer
I tried to follow these steps is not working
Certainly! Here are instructions on how to remove a personal email from your computer:
1. Windows Settings Method:
o Go to Start > Settings > Accounts > Email & accounts.
o Sign out of your personal account if it’s listed there.
o If you don’t see a “Remove” button, click on “Manage” instead.
o Log into your Microsoft account under “Device & activity.”
o Disable this work computer from accessing your email account.
before we use to have the option to disconnect the email address but now is saying manage and when clicked on manage nothing comes out of it.
Hello Please i need your help on this issue. There is no option to remove personal email from the computerI tried to follow these steps is not workingCertainly! Here are instructions on how to remove a personal email from your computer:1. Windows Settings Method:o Go to Start > Settings > Accounts > Email & accounts.o Sign out of your personal account if it’s listed there.o If you don’t see a “Remove” button, click on “Manage” instead.o Log into your Microsoft account under “Device & activity.”o Disable this work computer from accessing your email account.before we use to have the option to disconnect the email address but now is saying manage and when clicked on manage nothing comes out of it. Read More
How we can force the user to sign in with their work account to use Bing Copilot.
How we can force the user to sign in with their work account to use Bing Copilot, due to security risks / DLP issues etc.
How we can force the user to sign in with their work account to use Bing Copilot, due to security risks / DLP issues etc. Read More
DataFormat.Error: We couldn’t parse the input provided as a Date value. Details: Unknown
Hi all, I’m using Power Query Editor in Excel to edit my data. I’m getting a DataFormat error when I try to expand a merged table if I merged on two columns. It doesn’t happen with any data if I merge on one column only, and always happens if I try to merge on two columns.
All my data is in its correct format, and it doesn’t matter if I’m trying to expand a date column or not (the column I’m trying to pull through is a character column).
Does anybody have any idea why this might be happening and how to prevent it?
Thanks
Hi all, I’m using Power Query Editor in Excel to edit my data. I’m getting a DataFormat error when I try to expand a merged table if I merged on two columns. It doesn’t happen with any data if I merge on one column only, and always happens if I try to merge on two columns. All my data is in its correct format, and it doesn’t matter if I’m trying to expand a date column or not (the column I’m trying to pull through is a character column). Does anybody have any idea why this might be happening and how to prevent it? Thanks Read More
How Microsoft is empowering organizations: News and updates from FinOps X 2024
Last year, I shared a broad set of updates that showcased how Microsoft is embracing FinOps practitioners through education, product improvements, and innovative solutions that help organizations achieve more. This year, at FinOps X, we shared the first steps in a transformative vision to help accelerate FinOps and empower stakeholders across the organization with AI-powered experiences like Copilot and Microsoft Fabric. Whether you’re an engineer working in the Azure portal or part of a business or finance team collaborating in Microsoft 365 or analyzing data in Power BI, Microsoft Cloud has the tools you need to accelerate business value for your cloud investments.
I’m about to take you on a whirlwind tour of updates to products, solutions, and services from across Microsoft over the past few months. Some of these were announced at FinOps X and some are new. But let me start with the biggest announcements, which I’ll touch on in more detail throughout the blog post.
Leverage new resources to learn about FinOps and supporting Microsoft tools and services.
Estimate costs before you deploy virtual machines.
Export cost and usage data using FOCUS 1.0 in Microsoft Cost Management.
Ingest FinOps data into Microsoft Fabric to empower business and finance with self-serve analytics.
Accelerate FinOps efforts with Microsoft Copilot in the Azure portal and Microsoft Fabric.
Of course, these are just teasers. Read on to learn more and don’t forget to check out the video below to see the power of Copilot and Microsoft Fabric in action!
Empowering engineers
Empowering people always starts with education, so to kick things off, I’d like to share new learning resources like the Get started with FinOps learning module for those new to FinOps and the Adopt FinOps on Azure learning module that introduces the FinOps capabilities and helps you get started. And once you’re ready to jump into the tools, check out the interactive guides for a walk through of the products and solutions you’ll need as you implement FinOps.
If you’re looking for more, talk to your Microsoft account team and leverage the FinOps offerings from Microsoft Services. As a FinOps Certified Service Provider, Microsoft offers a broad range of services that span every corner of FinOps and more, but just to give you an idea, here are some of the ones you might be interested in:
Strategic planning
FinOps assessment
FinOps strategy planning
FinOps operations
Educate and asses
Azure cost optimization
Azure cost optimization assessment
FinOps technical implementation
Technical implementation
FinOps technical implementation
Azure cost optimization
Next, let’s jump into the portal where you can now drill into the estimated cost of virtual machines before you deploy. This brings the power and transparency of the Azure pricing calculator directly into the Azure portal to give you a complete, upfront picture of your pay-as-you-go virtual machine costs. Stay tuned as this new experience gets rolled out to other services and account types.
And in case you missed the general availability announcement at KubeCon, you’ll also find new smart views for Azure Kubernetes Service (AKS) in Cost analysis where you can get more transparency and drill into AKS clusters to view idle, used, and system costs of each resource or the compute, network, and storage costs of each namespace. With these additional insights, engineers can optimize their AKS costs more efficiently, maximizing the benefits of running their workloads on shared infrastructure.
And speaking of optimization, there’s a ton to share, starting with new and updated tools in Azure Advisor to help you implement proven practices from the Well-Architected Framework (WAF). You’ll find updates to the Cost optimization workbook, which gives you a holistic view of your environment to optimize rates and reduce waste, and discover new well-architected and mission-critical assessments directly from Advisor. Each self-service assessment provides guidelines as recommendations, which can be managed and tracked over time. And if you’re looking for a more holistic view of FinOps, you can also leverage the FinOps review assessment.
You’ll also find a broad range of rate optimization updates, like configuring auto-renewal for reservations at purchase time and new RBAC roles for savings plans, but I’m especially excited about expanded savings plan and reservation offers. I wish I could share more about the new offers that are in the works, but you’ll have to stay tuned.
New reservation offers for:
SQL Database Hyperscale (2 SKUs)
SQL Managed Instance (4 SKUs)
Microsoft Fabric
Microsoft Defender for Cloud
Savings Plan for Compute adds support for:
Azure Container Apps
Azure Spring Apps
But FinOps isn’t just about cost. With the expansion of the FinOps Framework to also cover cloud sustainability, I have to call out the preview release of Azure Carbon Optimization earlier this year, where engineers can monitor, track, and analyze emission trends at the resource level from the Azure portal. And with efficiency recommendations that help optimize carbon emissions, Azure Carbon Optimization provides the foundational tools you need to get started on your cloud sustainability journey.
And whether you’re learning about or looking for solutions to automate and implement FinOps capabilities in the Microsoft Cloud, the FinOps toolkit helps kick-start your FinOps journey with starter kits, scripts, and more. Next week you’ll find new and updated tools, like the Azure Optimization Engine for custom recommendations, updates to the FinOps hubs data ingestion engine to support managed exports and organizations with multi-tenant environments, and updates to the Implementing FinOps guide to align with the FinOps Framework 2024 updates.
Empowering business and finance
Looking beyond engineers, we’re also exploring new ways to empower business and finance teams in the tools they use outside of the Azure portal.
And the first step is getting Microsoft Cloud data into these tools through capabilities like Cost Management exports, which offer no-code data ingestion with enhanced security and 5 new datasets, making it easier to get the data you need into a central reporting tool. The exports preview was first launched in November and we’re happy to share a streamlined create experience where you can select a template covering all the datasets you need as well as an extended self-service data export for the past 7 years. You can export data for the last 13 months from the Azure portal. To go back further, you can use the Exports Execute API or the Start-FinOpsCostExport PowerShell command in the FinOps toolkit.
And perhaps one of the most exciting and anticipated datasets you can export is cost and usage data aligned to the FinOps Open Cost and Usage Specification (FOCUS) schema. As the first cloud provider to implement FOCUS in November 2023, we’ve seen incredible adoption and interest. Organizations of all shapes and sizes are exporting FOCUS data and migrating their reports and alerts. As one of the key leaders in the development of FOCUS, seeing this has been inspiring. And now, you can also export cost and usage data with FOCUS 1.0 to get the latest specification enhancements. Along with this, the next release of the FinOps toolkit will include updated Power BI reports to support your FOCUS 1.0 exports.
And to truly democratize FinOps and empower business and finance teams, leverage Microsoft Fabric to enable self-serve analytics, connect to your existing internal data sources, and streamline some of the most complex FinOps reports and alerts on top of FOCUS 1.0 and other datasets, including carbon emissions. By linking Cost Management exports and enabling Microsoft Azure Emissions Insights in Microsoft Fabric, you can tell the full FinOps story, from cost to usage to carbon and tying it all back to business value to quantify cloud ROI by leveraging your existing business data, like application telemetry.
Accelerating FinOps with AI
And of course, it wouldn’t be a Microsoft update without AI :beaming_face_with_smiling_eyes: But instead of walking you through it all, I’ll share a short video which demonstrates what you can do today with Microsoft Copilot in the Azure portal or in Microsoft Fabric, powered by both your cloud and business data joined together in one powerful place.
Next steps
I hope you’re as excited as I am about the new capabilities available in the Azure portal and the potential of what you can do with Microsoft Fabric. We first hinted at the potential Microsoft Fabric could bring in November, but this is just the beginning. I’ve never been as excited about what the future holds as I am today. Stay tuned for more updates on the FinOps blog.
Microsoft Tech Community – Latest Blogs –Read More
Fine-tuning Florence-2 for VQA (Visual Question Answering) using the Azure ML Python SDK/and MLflow
Released by Microsoft in mid-June 2024 under an MIT license, Florence-2 is less than 1B in size (0.23B for the base model and 0.77B for the large model) and is efficient for vision and vision-language tasks (OCR, captioning, object detection, instance segmentation, and so on).
All of Florence-2’s weights are publicly available, so you can fine-tune it quickly and easily. However, many people struggle with fine-tuning the latest SLM/multi-modal models, including Florence-2, in Azure ML studio. So, we want to walk through a step-by-step guide on how to quickly and easily train and serve from end-to-end in Azure ML.
1. Training preparation
1.1. Preliminaries: Azure ML Python SDK v2
Azure ML Python SDK v2 is easy to use once you get the hang of it. When an MLClient instance is created to manipulate AzureML, the operation corresponding to the asset is executed asynchronously through the create_or_update function. Please see code snippets below.
1.2. Data asset
Model training/validation datasets can be uploaded directly locally, or registered as your Azure ML workspace Data asset. Data asset enables versioning of your data, allowing you to track changes to your dataset and revert to previous versions when necessary. This maintains data quality and ensures reproducibility of data analysis.
Data assets are created by referencing data files or directories stored in Datastore. Datastore represents a location that stores external data and can be connected to various Azure data storage services such as Azure Blob Storage, Azure File Share, Azure Data Lake Storage, and OneLake. When you create an Azure ML workspace, four datastores (workspaceworkingdirectory, workspaceartifactstore, workspacefilestore, workspaceblobstore) are automatically created by default. Among these, workspaceblobstore is Azure Blob Storage, which is used by default when storing model training data or large files.
1.3. Environment asset
Azure ML defines Environment Asset in which your code will run. We can use the built-in environment or build a custom environment using Conda specification or Docker image. The pros and cons of Conda and Docker are as follows.
Conda environment
Advantages
Simple environment setup: The Conda environment file (conda.yml) is mainly used to specify Python packages and Conda packages. The file format is simple and easy to understand, and is suitable for specifying package and version information.
Quick setup: The Conda environment automatically manages dependencies and resolves conflicts, so setup is relatively quick and easy.
Lightweight environment: Conda environments can be lighter than Docker images because they only install specific packages.
Disadvantages
Limited flexibility: Because the Conda environment focuses on Python packages and Conda packages, it is difficult to handle more complex system-level dependencies.
Portability limitations: The Conda environment consists primarily of Python and Conda packages, making it difficult to include other languages or more complex system components.
Docker environment
Advantages
High flexibility: Docker allows you to define a complete environment, including all necessary packages and tools, starting at the operating system level. May contain system dependencies, custom settings, non-Python packages, etc.
Portability: Docker images run the same everywhere, ensuring environment consistency. This significantly improves reproducibility and portability.
Complex environment setup: With Docker, you can set up an environment containing complex applications or multiple services.
Disadvantages
Complex setup: Building and managing Docker images can be more complex than setting up a Conda environment. You need to write a Dockerfile and include all required dependencies.
Build time: Building a Docker image for the first time can take a long time, especially if the dependency installation process is complex.
In Azure ML, it is important to choose the appropriate method based on the requirements of your project. For simple Python projects, the Conda environment may be sufficient, but if you need complex system dependencies, the Docker environment may be more appropriate. The easiest and fastest way to create a custom Docker image is to make minor modifications to the curated environment. Below is an example.
Select acft-hf-nlp-gpu in the cured environment tab. (Of course, you can choose a different environment.)
Copy the Dockerfile and requirements.txt and modify them as needed.
The code snippet below is the result of modifying the Dockerfile.
FROM mcr.microsoft.com/aifx/acpt/stable-ubuntu2004-cu118-py38-torch222:biweekly.202406.2
USER root
RUN apt-get update && apt-get -y upgrade
RUN pip install –upgrade pip
COPY requirements.txt .
RUN pip install -r requirements.txt –no-cache-dir
RUN python -m nltk.downloader punkt
RUN MAX_JOBS=4 pip install flash-attn==2.5.9.post1 –no-build-isolation
2. Training
2.1. Training Script with MLflow
Some people may think that they need to make significant changes to their existing training scripts or that the Mlflow toolkit is mandatory, but this is not true. If you are comfortable with your existing training environment, you don’t need to adopt Mlflow. Nevertheless, Mlflow is a toolkit that makes training and deploying models on Azure ML very convenient, so we are going to briefly explain it in this post.
In the your training script, Use mlflow.start_run() to start an experiment in MLflow, and mlflow.end_run() to end the experiment when it is finished. Wrapping it in with syntax eliminates the need to explicitly call end_run(). You can perform mlflow logging inside an mlflow block, our training script uses mlflow.log_params(), mlflow.log_metric(), and mlflow.log_image(). For more information, please see here.
import mlflow
…
with mlflow.start_run() as run:
mlflow.log_params({
“epochs”: epochs,
“train_batch_size”: args.train_batch_size,
“eval_batch_size”: args.eval_batch_size,
“seed”: args.seed,
“lr_scheduler_type”: args.lr_scheduler_type,
“grad_accum_steps”: grad_accum_steps,
“num_training_steps”: num_training_steps,
“num_warmup_steps”: num_warmup_steps,
})
# Your training code
for epoch in range(epochs):
train_loss = 0.0
optimizer.zero_grad()
for step, (inputs, answers) in enumerate(tqdm(train_loader, desc=f”Training Epoch {epoch + 1}/{epochs}”)):
…
mlflow.log_metric(“train_loss”, train_loss)
mlflow.log_metric(“learning_rate”, learning_rate)
mlflow.log_metric(“progress”, progress)
…
if (step + 1) % save_steps == 0:
# Log image
idx = random.randrange(len(val_dataset))
val_img = val_dataset[idx][-1]
result = run_example(“DocVQA”, ‘What do you see in this image?’, val_dataset[idx][-1])
val_img_result = create_image_with_text(val_img, json.dumps(result))
mlflow.log_image(val_img_result, key=”DocVQA”, step=step)
[Caution] Florence-2 is a recently released model and does not support mlflow.transformers.log_model() as of July 2, 2024, when this article is being written! Therefore, you must save the model with the traditional save_pretrained().
Currently, when saving with save_pretrained(), additional dependency codes required for model inference are not saved together. So, you need to force it to be saved. See below for a code snippet reflecting these two caveats.
model.save_pretrained(model_dir)
processor.save_pretrained(model_dir)
## Should include configuration_florence2.py, modeling_florence2.py, and processing_florence2.py
dependencies_dir = “dependencies”
shutil.copytree(dependencies_dir, model_dir, dirs_exist_ok=True)
2.2. Create a Compute Cluster and Training Job
Once you have finished writing and debugging the training script, you can create a training job. As a baseline, you can use Standard_NC24ads_A100_v4 with one NVIDIA A100 GPU. Provisioning a LowPriority VM costs just $0.74 per hour in the US East region in July 2024.
The command() function is one of the Azure ML main functions used to define and run training tasks. This function specifies the training script and its required environment settings, and allows the job to be run on Azure ML’s compute resources.
from azure.ai.ml import command
from azure.ai.ml import Input
from azure.ai.ml.entities import ResourceConfiguration
job = command(
inputs=dict(
#train_dir=Input(type=”uri_folder”, path=DATA_DIR), # Get data from local path
train_dir=Input(path=f”{AZURE_DATA_NAME}@latest”), # Get data from Data asset
epoch=d[‘train’][‘epoch’],
train_batch_size=d[‘train’][‘train_batch_size’],
eval_batch_size=d[‘train’][‘eval_batch_size’],
model_dir=d[‘train’][‘model_dir’]
),
code=”./src_train”, # local path where the code is stored
compute=azure_compute_cluster_name,
command=”python train_mlflow.py –train_dir ${{inputs.train_dir}} –epochs ${{inputs.epoch}} –train_batch_size ${{inputs.train_batch_size}} –eval_batch_size ${{inputs.eval_batch_size}} –model_dir ${{inputs.model_dir}}”,
#environment=”azureml://registries/azureml/environments/acft-hf-nlp-gpu/versions/61″, # Use built-in Environment asset
environment=f”{azure_env_name}@latest”,
distribution={
“type”: “PyTorch”,
“process_count_per_instance”: 1, # For multi-gpu training set this to an integer value more than 1
},
)
returned_job = ml_client.jobs.create_or_update(job)
ml_client.jobs.stream(returned_job.name)
2.3. Check your Training job
Check whether model training is progressing normally through Jobs Asset.
Overview tab allows you to view your overall training history. Params are parameters registered in mlflow.log_params() in our training script.
Metrics tab allows you to view the metrics registered with mlflow.log_metric() at a glance.
Images tab allows you to view images saved with mlflow.log_image(). We recommend that you save the inference results as an image to check whether the model training is progressing well.
Outputs + logs tab checks and monitors your model training infrastructure, containers, and code for issues.
system_logs folder records all key activities and events related to the Training cluster, data assets, hosted tools, etc.
user_logs folder mainly plays an important role in storing logs and other files created by users within the training script, increasing transparency of the training process and facilitating debugging and monitoring. This allows users to see a detailed record of the training process and identify and resolve issues when necessary.
3. Serving
Once the model training is complete, let’s deploy it to the hosting server. If you saved it with MLflow log_model(), you can deploy it directly with Mlflow, but in the current transformer and mlflow version, we used the traditional way of saving the model, so we need to deploy it with the custom option.
3.1. Inference script
You only need to define two functions, init() and run(), and write them freely. Although you cannot pass arguments to the init() function directly, you can pass the necessary information during initialization through environment variables or configuration files.
import os
import re
import json
import torch
import base64
import logging
from io import BytesIO
from PIL import Image
from transformers import AutoTokenizer, AutoProcessor, BitsAndBytesConfig, get_scheduler
from transformers import AutoModelForCausalLM, AutoProcessor
from PIL import Image, ImageDraw, ImageFont
device = torch.device(“cuda” if torch.cuda.is_available() else “cpu”)
def run_example_base64(task_prompt, text_input, base64_image, params):
max_new_tokens = params[“max_new_tokens”]
num_beams = params[“num_beams”]
image = Image.open(BytesIO(base64.b64decode(base64_image)))
prompt = task_prompt + text_input
# Ensure the image is in RGB mode
if image.mode != “RGB”:
image = image.convert(“RGB”)
inputs = processor(text=prompt, images=image, return_tensors=”pt”).to(device)
generated_ids = model.generate(
input_ids=inputs[“input_ids”],
pixel_values=inputs[“pixel_values”],
max_new_tokens=max_new_tokens,
num_beams=num_beams
)
generated_text = processor.batch_decode(generated_ids, skip_special_tokens=False)[0]
parsed_answer = processor.post_process_generation(generated_text, task=task_prompt, image_size=(image.width, image.height))
return parsed_answer
def init():
“””
This function is called when the container is initialized/started, typically after create/update of the deployment.
You can write the logic here to perform init operations like caching the model in memory
“””
global model
global processor
# AZUREML_MODEL_DIR is an environment variable created during deployment.
# It is the path to the model folder (./azureml-models/$MODEL_NAME/$VERSION)
# Please provide your model’s folder name if there is one
model_name_or_path = os.path.join(
os.getenv(“AZUREML_MODEL_DIR”), “outputs”
)
model_kwargs = dict(
trust_remote_code=True,
revision=”refs/pr/6″,
device_map=device
)
processor_kwargs = dict(
trust_remote_code=True,
revision=”refs/pr/6″
)
model = AutoModelForCausalLM.from_pretrained(model_name_or_path, **model_kwargs)
processor = AutoProcessor.from_pretrained(model_name_or_path, **processor_kwargs)
logging.info(“Loaded model.”)
def run(json_data: str):
logging.info(“Request received”)
data = json.loads(json_data)
task_prompt = data[“task_prompt”]
text_input = data[“text_input”]
base64_image = data[“image_input”]
params = data[‘params’]
generated_text = run_example_base64(task_prompt, text_input, base64_image, params)
json_result = {“result”: str(generated_text)}
return json_result
3.2. Register Model
Register with the Model class of azure.ai.ml.entities. Enter the model’s path and name when registering and use with ml_client.models.create_or_update().
def get_or_create_model_asset(ml_client, model_name, job_name, model_dir=”outputs”, model_type=”custom_model”, update=False):
try:
latest_model_version = max([int(m.version) for m in ml_client.models.list(name=model_name)])
if update:
raise ResourceExistsError(‘Found Model asset, but will update the Model.’)
else:
model_asset = ml_client.models.get(name=model_name, version=latest_model_version)
print(f”Found Model asset: {model_name}. Will not create again”)
except (ResourceNotFoundError, ResourceExistsError) as e:
print(f”Exception: {e}”)
model_path = f”azureml://jobs/{job_name}/outputs/artifacts/paths/{model_dir}/”
run_model = Model(
name=model_name,
path=model_path,
description=”Model created from run.”,
type=model_type # mlflow_model, custom_model, triton_model
)
model_asset = ml_client.models.create_or_update(run_model)
print(f”Created Model asset: {model_name}”)
return model_asset
3.3. Environment asset
This is the same as the Environment asset introduced in the previous section. However, model serving requires additional settings for web hosting, so please refer to the code snippet below.
FROM mcr.microsoft.com/aifx/acpt/stable-ubuntu2004-cu118-py38-torch222:biweekly.202406.2
# Install pip dependencies
COPY requirements.txt .
RUN pip install -r requirements.txt –no-cache-dir
RUN MAX_JOBS=4 pip install flash-attn==2.5.9.post1 –no-build-isolation
# Inference requirements
COPY –from=mcr.microsoft.com/azureml/o16n-base/python-assets:20230419.v1 /artifacts /var/
RUN /var/requirements/install_system_requirements.sh && \
cp /var/configuration/rsyslog.conf /etc/rsyslog.conf && \
cp /var/configuration/nginx.conf /etc/nginx/sites-available/app && \
ln -sf /etc/nginx/sites-available/app /etc/nginx/sites-enabled/app && \
rm -f /etc/nginx/sites-enabled/default
ENV SVDIR=/var/runit
ENV WORKER_TIMEOUT=400
EXPOSE 5001 8883 8888
# support Deepspeed launcher requirement of passwordless ssh login
RUN apt-get update
RUN apt-get install -y openssh-server openssh-client
3.4. Create an Endpoint
An endpoint refers to an HTTP(S) URL that makes the model accessible from the outside. Endpoint can have multiple deployments, which allows traffic to be distributed across multiple deployments. Endpoint does the following:
API interface provided: Endpoint provides a URL to receive model prediction requests through a RESTful API.
Traffic routing: Endpoint distributes traffic across multiple deployments. This allows you to implement A/B testing or canary deployment strategies.
Scalability: Endpoint supports scaling across multiple deployments and can be load balanced across additional deployments as traffic increases.
Security Management: Endpoints secure models through authentication and authorization. You can control access using API keys or Microsoft Entra ID.
The code snippet is below. Note that this process does not provision a compute cluster yet.
from azure.ai.ml.entities import (
ManagedOnlineEndpoint,
IdentityConfiguration,
ManagedIdentityConfiguration,
)
# Check if the endpoint already exists in the workspace
try:
endpoint = ml_client.online_endpoints.get(azure_endpoint_name)
print(“—Endpoint already exists—“)
except:
# Create an online endpoint if it doesn’t exist
endpoint = ManagedOnlineEndpoint(
name=azure_endpoint_name,
description=f”Test endpoint for {model.name}”,
)
# Trigger the endpoint creation
try:
ml_client.begin_create_or_update(endpoint).wait()
print(“\n—Endpoint created successfully—\n”)
except Exception as err:
raise RuntimeError(
f”Endpoint creation failed. Detailed Response:\n{err}”
) from err
3.5. Create a Deployment
Deployment is the instance that actually run the model. Multiple deployments can be connected to an endpoint, and each deployment contains a model, environment, compute resources, infrastructure settings, and more. Deployment does the following:
Manage resources: The deployment manages the computing resources needed to run the model. You can set up resources like CPU, GPU, and memory.
Versioning: Deployments can manage different versions of a model. This makes it easy to roll back to a previous version or deploy a new version.
Monitoring and logging: We can monitor the logs and performance of running models. This helps you detect and resolve issues.
The code snippet is below. Note that this takes a lot of time as a GPU cluster must be provisioned and the serving environment must be built.
from azure.ai.ml.entities import (
OnlineRequestSettings,
CodeConfiguration,
ManagedOnlineDeployment,
ProbeSettings,
Environment
)
deployment = ManagedOnlineDeployment(
name=azure_deployment_name,
endpoint_name=azure_endpoint_name,
model=model,
instance_type=azure_serving_cluster_size,
instance_count=1,
#code_configuration=code_configuration,
environment = env,
scoring_script=”score.py”,
code_path=”./src_serve”,
#environment_variables=deployment_env_vars,
request_settings=OnlineRequestSettings(max_concurrent_requests_per_instance=3,
request_timeout_ms=90000, max_queue_wait_ms=60000),
liveness_probe=ProbeSettings(
failure_threshold=30,
success_threshold=1,
period=100,
initial_delay=500,
),
readiness_probe=ProbeSettings(
failure_threshold=30,
success_threshold=1,
period=100,
initial_delay=500,
),
)
# Trigger the deployment creation
try:
ml_client.begin_create_or_update(deployment).wait()
print(“\n—Deployment created successfully—\n”)
except Exception as err:
raise RuntimeError(
f”Deployment creation failed. Detailed Response:\n{err}”
) from err
endpoint.traffic = {azure_deployment_name: 100}
endpoint_poller = ml_client.online_endpoints.begin_create_or_update(endpoint)
[Tip] Please specify and deploy the liveness probe settings directly to check if the model deployment container is running normally. When debugging, it is recommended to set a high initial_delay and a high failure_threshold and high period for error log analysis. Please check ProbeSettings() in the code above.
4. Invocation
We finally succeeded in serving the Florence-2 model. Try using the code below to perform model inference.
import os
import json
import base64
with open(‘./DocumentVQA_Test_01.jpg’, ‘rb’) as img:
base64_img = base64.b64encode(img.read()).decode(‘utf-8’)
sample = {
“task_prompt”: “DocVQA”,
“image_input”: base64_img,
“text_input”: “What do you see in this image”,
“params”: {
“max_new_tokens”: 512,
“num_beams”: 3
}
}
test_src_dir = “./inference-test”
os.makedirs(test_src_dir, exist_ok=True)
print(f”test script directory: {test_src_dir}”)
sample_data_path = os.path.join(test_src_dir, “sample-request.json”)
with open(sample_data_path, “w”) as f:
json.dump(sample, f)
result = ml_client.online_endpoints.invoke(
endpoint_name=azure_endpoint_name,
deployment_name=azure_deployment_name,
request_file=sample_data_path,
)
result_json = json.loads(result)
print(result_json[‘result’])
It is a good strategy to perform LLM latency/throughput benchmarking before deploying the model in earnest. Benchmark the following metrics as a baseline.
metrics = {
‘threads’: num_threads,
‘duration’: duration,
‘throughput’: throughput,
‘avg_sec’: avg_latency,
‘std_sec’: time_std_sec,
‘p95_sec’: time_p95_sec,
‘p99_sec’: time_p99_sec
}
We have published the code to do this post end-to-end at https://github.com/Azure/azure-llm-fine-tuning/tree/main/florence2-VQA.
We hope this tutorial will help you fine-tune and deploy modern models, including the Florence-2 model, in Azure ML Studio.
References
Hugging Face blog – Fine-tuning Florence-2
Fine-tune SLM Phi-3 using Azure ML
Hands-on labs – LLM Fine-tuning/serving with Azure ML
Microsoft Tech Community – Latest Blogs –Read More
Example of Powershell Script to creat CQ and AA
Hello
I am looking for a Website that has allready a script to create CQ and AA with a CSV.
The usecase would be
1. Create CQ
2. Create AA
3. Assign phone number to AA
4. Connect both of them together
5. Create Teams team
6. Connect CQ with Teams
or
1. Do the above with CQ only without AA
2. Do the above with AA only without CQ
Do you have a website that has this.
Regards
JFM_12
HelloI am looking for a Website that has allready a script to create CQ and AA with a CSV. The usecase would be1. Create CQ2. Create AA3. Assign phone number to AA4. Connect both of them together5. Create Teams team6. Connect CQ with Teamsor 1. Do the above with CQ only without AA2. Do the above with AA only without CQ Do you have a website that has this. Regards JFM_12 Read More
Why is Quick-Books Not Working on Windows and How Can I Fix It?
I’m encountering issues with Q.B not working properly on my Windows computer. Whenever I attempt to open the program, it crashes or freezes, preventing me from accessing my financial data. I’ve already tried basic troubleshooting steps like restarting my computer and reinstalling the software, but the problem persists. Can someone provide detailed troubleshooting steps or solutions specific to Windows to help me resolve this issue? Any assistance would be greatly appreciated.
I’m encountering issues with Q.B not working properly on my Windows computer. Whenever I attempt to open the program, it crashes or freezes, preventing me from accessing my financial data. I’ve already tried basic troubleshooting steps like restarting my computer and reinstalling the software, but the problem persists. Can someone provide detailed troubleshooting steps or solutions specific to Windows to help me resolve this issue? Any assistance would be greatly appreciated. Read More
Office 365 apps are closing randomly on MacOS
Hi,
is anyone currently experiencing issues with all MS Office 365 apps (Outlook, Excel, Powerpoint, OneNote, Word) in the way that they close all at once, randomly, and you are probably losing work due to the unexpected shutdown.
I got a new MacBook Pro with M3 Max with Sonoma 14.5 recently, and after using it for a few hours, it seems that sporadically, all O365 apps mentioned above that are open at the moment are closing all at once, randomly, without any error message, without any prior notice or any user interaction. ALL other applications except Office are working fine, also OneDrive and MS Teams stay open without any issues. After using the MacBook for a few days, it seems that it’s maybe more likely to happen when the Mac is going to standby / lid closed and is woken up afterwards?
Steps I already tried without any change of the behavior, each also including a complete reboot of the machine, in the following order.
Updating Office with the MS AutoUpdater applicationSimple uninstall of MS Office applicationsReinstallation by using a clean new O365 downloadManual uninstallation using https://support.microsoft.com/en-us/office/uninstall-office-for-mac-eefa1199-5b58-43af-8a3d-b73dc1a8cae3Clean reinstallation after complete manual uninstallationAfter uninstalling again, I tried to install it manually with deselection of the MS Defender which is included in the O365 installer package (Defender is already installed by default on the Mac).Renaming the MacBooks hostname from XXX-MBP-ABC123DEF to XXXMBPABC123DEF and renaming the SSD name from Macintosh HD to MacintoshHD. Complete wipe of the MacBook with my IT department and reinstall / setup of the machine, followed by starting auto updater and updating everything to the latest version.
None of these steps is working, and Office keeps shutting down / crashing without any prior notice at random times, most likely after a sleep. It is enough to just open up some office applications and leave the Mac alone, after return you will find the Mac with all office applications closed (except Teams and Onedrive as mentioned above).
Furthermore, I already set the network availability during sleep within my energy saving settings:
Wake for network access from “Only on Power adapter” to “Always”
Does anyone have any further ideas for analysis or a solution?
On my private MBP with M1 Pro and Sonoma 14.5 there are no issues at all.
Thank you!!!
Hi,is anyone currently experiencing issues with all MS Office 365 apps (Outlook, Excel, Powerpoint, OneNote, Word) in the way that they close all at once, randomly, and you are probably losing work due to the unexpected shutdown. I got a new MacBook Pro with M3 Max with Sonoma 14.5 recently, and after using it for a few hours, it seems that sporadically, all O365 apps mentioned above that are open at the moment are closing all at once, randomly, without any error message, without any prior notice or any user interaction. ALL other applications except Office are working fine, also OneDrive and MS Teams stay open without any issues. After using the MacBook for a few days, it seems that it’s maybe more likely to happen when the Mac is going to standby / lid closed and is woken up afterwards? Steps I already tried without any change of the behavior, each also including a complete reboot of the machine, in the following order.Updating Office with the MS AutoUpdater applicationSimple uninstall of MS Office applicationsReinstallation by using a clean new O365 downloadManual uninstallation using https://support.microsoft.com/en-us/office/uninstall-office-for-mac-eefa1199-5b58-43af-8a3d-b73dc1a8cae3Clean reinstallation after complete manual uninstallationAfter uninstalling again, I tried to install it manually with deselection of the MS Defender which is included in the O365 installer package (Defender is already installed by default on the Mac).Renaming the MacBooks hostname from XXX-MBP-ABC123DEF to XXXMBPABC123DEF and renaming the SSD name from Macintosh HD to MacintoshHD. Complete wipe of the MacBook with my IT department and reinstall / setup of the machine, followed by starting auto updater and updating everything to the latest version.None of these steps is working, and Office keeps shutting down / crashing without any prior notice at random times, most likely after a sleep. It is enough to just open up some office applications and leave the Mac alone, after return you will find the Mac with all office applications closed (except Teams and Onedrive as mentioned above).Furthermore, I already set the network availability during sleep within my energy saving settings:Wake for network access from “Only on Power adapter” to “Always” Does anyone have any further ideas for analysis or a solution?On my private MBP with M1 Pro and Sonoma 14.5 there are no issues at all. Thank you!!! Read More