Month: August 2024
SharePoint Online > Sites > Deleted Sites – Column defeinitions
Hello all,
I am looking at the SharePoint Admin Center > Sites > Deleted Sites page
And there are multiple columns available, including Site name, URL, Storage used (GB) just to name a few.
But I am confused by the Date Created and Time Deleted columns, they do not appear to coordinate with what I know as the dates and times these sites were created and/or deleted. I know these columns are likely deemed self-explanatory, but again, I know that the dates and times shown are not accurate.
I’ve been searching the web for a couple days and all I’m finding are instructions on how to restore a deleted site, perhaps my Google-Fu is failing terribly, but is there any documentation concerning this page that includes the column definitions and how their data is derived?
Thanks in advance,
-David Covert
Hello all, I am looking at the SharePoint Admin Center > Sites > Deleted Sites pageAnd there are multiple columns available, including Site name, URL, Storage used (GB) just to name a few.But I am confused by the Date Created and Time Deleted columns, they do not appear to coordinate with what I know as the dates and times these sites were created and/or deleted. I know these columns are likely deemed self-explanatory, but again, I know that the dates and times shown are not accurate. I’ve been searching the web for a couple days and all I’m finding are instructions on how to restore a deleted site, perhaps my Google-Fu is failing terribly, but is there any documentation concerning this page that includes the column definitions and how their data is derived? Thanks in advance, -David Covert Read More
Server 2025 Hotpatching | Windows Azure Arc VMs
Hello!
I was trying to test Hotpatching with an on-prem Server 2025 Standard. It has Azure Arc installed on it, but when I tried to apply a Hotpatching policy in Azure, it said the VM was not supported.
On-prem VMs with Azure Arc will have Hotpatching Support, correct?
Hello! I was trying to test Hotpatching with an on-prem Server 2025 Standard. It has Azure Arc installed on it, but when I tried to apply a Hotpatching policy in Azure, it said the VM was not supported. On-prem VMs with Azure Arc will have Hotpatching Support, correct? Read More
How To Reference Files Using Copilot In Word By Copy Pasting The URL
First off, I have no idea why what I’m about to type is not published in the existing Copilot in Word training materials produced by Microsoft. Nevertheless, if you are trying to use the pre-fetched files or trying to type in file names and they aren’t showing up….here is how to get around all of that. Also, it does require some pre-work in step number 1 – but this is a one time deal.
Search for clipboard settings in your Windows settingsTap the windows button on your keyboard and type clipboard…select “clipboard settings”Turn on “Clipboard History” to “On” if it isn’t on alreadyNavigate to the location of the file you want to reference (OneDrive/SharePoint)Open the details pane for the file and scroll down to the bottom to the “Path” section and click the icon. This copies the absolute URL (just the straight up path – no garbage included) to the fileRepeat as necessary until you have up to 3 file paths copiedGo to your Copilot prompt and type what you want to happen with the files you are about to reference from your clipboard (summarize, rewrite, create new file based on…..)Within that same prompt, hold down the windows button and press “V”This will bring up your clipboard history (if you followed the 1st step above correctly)Now you can paste in the URLs of the files you want to referenceYou will know the files were found/accepted after you hit the space bar as the URL will resolve into a hyperlink with the title of the file3 files is currently the max you can reference in one prompt
Yes this is a bit manual – however I see no other way to reference a file if it doesn’t show up in the prefetch window or when you start typing the name of the file.
Good luck and please let us know if this works for you!
First off, I have no idea why what I’m about to type is not published in the existing Copilot in Word training materials produced by Microsoft. Nevertheless, if you are trying to use the pre-fetched files or trying to type in file names and they aren’t showing up….here is how to get around all of that. Also, it does require some pre-work in step number 1 – but this is a one time deal. Search for clipboard settings in your Windows settingsTap the windows button on your keyboard and type clipboard…select “clipboard settings”Turn on “Clipboard History” to “On” if it isn’t on alreadyNavigate to the location of the file you want to reference (OneDrive/SharePoint)Open the details pane for the file and scroll down to the bottom to the “Path” section and click the icon. This copies the absolute URL (just the straight up path – no garbage included) to the fileRepeat as necessary until you have up to 3 file paths copiedGo to your Copilot prompt and type what you want to happen with the files you are about to reference from your clipboard (summarize, rewrite, create new file based on…..)Within that same prompt, hold down the windows button and press “V”This will bring up your clipboard history (if you followed the 1st step above correctly)Now you can paste in the URLs of the files you want to referenceYou will know the files were found/accepted after you hit the space bar as the URL will resolve into a hyperlink with the title of the file3 files is currently the max you can reference in one promptYes this is a bit manual – however I see no other way to reference a file if it doesn’t show up in the prefetch window or when you start typing the name of the file. Good luck and please let us know if this works for you! Read More
Elastic Logs sent for long term storage in Azure
Hello all,
I am using Elastic SIEM in my environment, but due to some pressing requirements – we would like to send the logs for long term storage. Now, I am not really sure how to send the Elastic SIEM logs to Azure Data Explorer (ADX), or to use Logic App to send the Elastic Logs to a Blob Storage.
Can you please help with both options? Is there any risk of losing data when using both options? which one is more feasible? Currently ingesting about 5 gigs of data in Elastic.
Thanks.
Hello all,I am using Elastic SIEM in my environment, but due to some pressing requirements – we would like to send the logs for long term storage. Now, I am not really sure how to send the Elastic SIEM logs to Azure Data Explorer (ADX), or to use Logic App to send the Elastic Logs to a Blob Storage.Can you please help with both options? Is there any risk of losing data when using both options? which one is more feasible? Currently ingesting about 5 gigs of data in Elastic. Thanks. Read More
How to find OneDrive users not running Sync
1. Introduction
We recently introduced a OneDrive Sync Health dataset in Microsoft Graph Data Connect. This provides a list of all devices running Sync, including details about whether they have enabled Folder Backup and are running into any errors. This ensures that any files being created or updated are properly synchronized with the OneDrive cloud. This makes sure that even if your computer is lost or broken, your files are safely stored.
However, there is another potential issue you should track as a OneDrive administrator. There could be users that are not running OneDrive Sync at all. These users could be saving files locally to their desktops or laptops without enabling OneDrive Sync. If their computer runs into any issues, they might find that their OneDrive in the cloud was not being updated at all.
This article explains how you can use the MGDC to get data about OneDrive sites and Sync Health to identify users that are not properly running Sync.
2. Grab the datasets
To get started, you need to get the two required MGDC datasets: Sites and Sync Health. To pull them, follow the instructions below:
Step-by-step: SharePoint Sites
Step-by-step: OneDrive Sync Health
The Sites dataset includes one object for each SharePoint site in the tenant. That includes every OneDrive, which is essentially just a type of SharePoint site. You can identify a OneDrive site by the template id (OneDrive uses template id 21).
The Sync Health dataset includes one object for each device running OneDrive Sync. This includes any attempt to synchronize in the last 30 days. It is important to note that if the user never enabled OneDrive Sync or has not attempted to synchronize in the last 30 days, there will be no data for that user in the Sync Health dataset.
Note: Some of the instructions are the same for both datasets. You can re-use things like the application registration, Azure storage account, MGDC consent and some of the Synapse resources.
3. Load into Power BI
The last step in the step-by-step instructions above is to load the datasets into Power BI. For this step, make sure to use Power BI desktop and load the two datasets into the same PBIX file.
After you’re done, the “Data” section on the right should show the two datasets. I would suggest renaming the datasets to “Sites” and “SyncHealth” instead of keeping the default names you get after you load the JSON files.
To rename them, just collapse the two datasets on the “data” section on the right (to make it easier to see them), then right-click the dataset name and use the “Edit query” option.
4. Filter the Sites for OneDrive
While editing the query for the Sites dataset, you can add a filter to include only the sites that are of type OneDrive. This is basically a filter on the RootWeb.TemplateId column where it equals 21. Simply find the arrow next to that column to enable a filter.
5. Manage relationships
With both datasets loaded into Power BI, you can now add a relationship between them. In the relationship view, click the “Manage relationships” in the bar. Or you can right click the dataset set on the Data section and select “Manage Relationships”.
You want to create a one-to-many relationship between Sites and Sync Health, matching the Site’s Owner.Email with Sync Health’s UserEmail.
6. Visualize using a table
Finally, create the visualization to see which OneDrive sites have no matching device in the Sync Health relationship. I used a simple “Table” visualization, pulling columns from either Sites or SyncHealth. Now that we have a relationship between the two, Power BI knows how to match data coming from both sides.
Note that I have renamed some columns in the table. There is an option to “rename for this visual” if you right click a column under “Visualizations”.
Also make sure to select the option to “Show items with no data” when adding Sync Health columns. This will make sure you show the sites with no matching device in Sync Health.
Here’s what your table can look like:
7. Conclusion
I hope this post helped you understand how to find OneDrive users that are not running OneDrive Sync. For more information about these and other scenarios in Microsoft Graph Data Connect for SharePoint, see the overview post at https://aka.ms/SharePointData.
Microsoft Tech Community – Latest Blogs –Read More
Announcing Public Preview of Instance Mix on Virtual Machine Scale Sets
Today, we’re excited to announce that the ability to specify multiple different VM sizes in your Virtual Machine Scale Sets (VMSS) with Flexible Orchestration Mode with Instance Mix. This new capability allows you to specify up to five different VM sizes to use in your VMSS deployments. Not only does this simplify the management of your scale sets, but it also allows you to optimize your deployments for either price or capacity through allocation strategies.
Benefits
Single API call: In one API call, you can specify and deploy up to five VM sizes in your scale sets.
Optimized deployments: You can optimize for the available capacity in Azure regions/zones with CapacityOptimized for more confidence that you’ll get the capacity requested. For cost-sensitive customers who want to still utilize capacity signals, the LowestPrice allocation strategy takes both factors into account.
Automated allocation: By using allocation strategies, let Azure determine which of the specified VMs are most readily available and most cost efficient.
Easily scale with a mix of VM offerings: When using Spot Priority Mix, you can deploy a heterogeneous mix of VM sizes of both Spot and On-Demand pricing in your scale sets.
Simplified management: You can now have the scale set deploy multiple different VM sizes, as opposed to manually deploying different VMs into the scale set.
Attain needed capacity: By utilizing a diversified mix of VM sizes, you can tap into a greater pool of capacity, increasing your ability to get the capacity you need.
What’s new
VMSS Instance Mix introduces a few new properties to the VMSS API including:
skuProfile: an umbrella property for the list of vmSizes and allocationStrategy.
vmSizes: the list of VM sizes to be deployed.
allocationStrategy: the property specify their allocation strategy. There are currently two options, LowestPrice and CapacityOptimized
LowestPrice: when evaluating what VM size split to use, Azure will first look at the lowest priced VMs of the VM sizes specified and then look at the capacity of those VMs. It will try to deploy as many of the lowest priced VMs it can before moving to the next lowest priced VM size with available capacity.
CapacityOptimized: when evaluating what VM size split to use, Azure will look at only the underlying capacity available and will not take price into consideration when determining what SKU split to deploy.
VMSS Instance Mix calls will have a new section in the ARM template:
“skuProfile”: {
“vmSizes”: [
{
“name”: “Standard_D16s_v4”
},
{
“name”: “Standard_E16s_v5”
},
{
“name”: “Standard_D16s_v5”
}
],
“allocationStrategy”: “CapacityOptimized” | ”LowestPrice”
},
You can learn more about how Instance Mix works and enroll in the preview in the documentation.
Microsoft Tech Community – Latest Blogs –Read More
Skilling snack: Mobile device management in Microsoft Intune
You’ve asked how to manage mobile and bring-your-own-device (BYOD) scenarios. With mobile device management (MDM), of course! Device management is an important part of a larger security strategy, so we’ve collected resources to help you get started or improve with MDM. Learn about deployment and management of Windows and various mobile platforms, endpoint and app management, update management, and how to handle management and deployment as securely as possible.
Time to learn: 136 minutes
READ
Mobile device management for Windows overview
Start managing company security policies and business applications while maintaining user privacy on personal devices. Check the MDM security baseline for your Windows versions as well as Windows editions and licensing requirements for Windows built-in management.
(4 mins)
Windows + Server + Policies + Security + Enrollment + Management + Protocol + Entra + FAQ
READ
What does device management mean?
Microsoft Intune is designed for management of mobile devices and applications. Learn about its features, benefits, and capabilities to manage various platforms including Windows. Choose the management solution that’s right for you, in the cloud or on premises.
(4 mins)
Intune + MDM + MAM + Apps + Windows + Android + iOS + iPadOS + Linux + macOS + ConfigMgr + Co-management
READ
Deployment guide: Manage devices running Windows
Here’s how to use Microsoft Intune to protect and manage Windows apps and endpoints. First, check your prerequisites and plan for deployment. Then, create compliance policies, configure endpoint security and device settings, and set up authentication methods. Finally, deploy apps, enroll devices, and run remote actions. Don’t forget to check out the next steps and help for employees and students.
(15 mins)
Intune + Admin Roles + Compliance + Policies + Access + Security + Apps + Authentication + Entra ID + Defender
WATCH
Provide access to on-premises resources for mobile devices using Microsoft Tunnel
Do users at your organization need access to resources that are located on premises? Learn about Microsoft Tunnel, which provides a secure, simple, and zero-trust-based solution for your mobile devices. Watch the explanation of its architecture, installation, configuration, and deployment for iOS and Android.
(26 mins)
Access + Tunnel + Cloud + On-premises + iOS + Android + Security
Enroll mobile devices in Microsoft Intune
Enrollment guide: Enroll Android devices in Microsoft Intune (17 mins)
Learn about the multiple options for enrolling Android devices in Microsoft Intune. These include BYOD devices with a work profile, Android Enterprise corporate owned dedicated devices and corporate owned fully managed devices, Android Enterprise corporate owned work profile, Android Open Source Project, and Android device administrator. Read an overview of the administrator and user tasks for each option and download visual guides.
Enrollment guide: Enroll iOS/iPadOS devices in Microsoft Intune (30 mins)
Choose the best way to enroll iOS and iPadOS devices in Microsoft Intune. Learn the difference among automated device enrollment (ADE) (supervised), Apple Configurator enrollment, and user and device enrollment. Find a list of administrator and user tasks for each option in this official documentation.
Intune + Android + iOS + iPadOS + BYOD + Enterprise + AOSP + ADE + Company Portal
Manage apps with Microsoft Intune
Microsoft Intune app management (1 min)
Easily manage apps on company-managed devices, whether they are organization-owned or personal. This functionality is called mobile application management (MAM). Watch this brief introduction to the topic before diving deeper!
What is Microsoft Intune app management? (9 min)
Start with the basics of MAM here. With Microsoft Intune, you can publish, push, configure, secure, monitor, and update mobile apps across your organization. Whether or not you choose to couple it with device management, you’ll find a range of capabilities by platform, a get-started guide, and additional functionalities to consider.
Data protection for Windows MAM (6 min)
Protect corporate data on personal devices used for work. Learn about Conditional Access compliance and threat defense health. Then create important Intune app protection policies (APP) with this guidance. Note: Windows MAM is supported in government cloud environments.
MAM + MDM + LOB + Apps + M365 + iOS + Android + Policies + Compliance + Admin Center + ACP + APP + GCC High + DoD + Security + Conditional Access + MTD + Data Protection + Health Checks
BOOKMARK
Microsoft Intune prescriptive guidance library
Save this library of interactive tutorials to learn how to use Intune as your MDM solution. Learn how to configure Conditional Access, deploy and manage line-of-business (LOB) applications, set up Windows Autopilot for new device delivery, update Windows from the cloud, and enable Universal Print.
Intune + MDM + Security + Apps + Autopilot + Updates + Universal Print
WATCH
Balancing security and flexibility when implementing Windows Defender Application Control (WDAC)
Consider WDAC as part of your MDM strategy. It’s a more comprehensive solution than AppLocker. Check out the video and chat comments for tips on how you can convert AppLocker policies to WDAC policies and get started with WDAC today.
(19 mins)
WDAC + Intune + Defender + MDE + App + Security + Allowlisting + Policies + Managed Installer
READ
Deploy WDAC policies using mobile device management (MDM)
If you use Microsoft Intune, configure WDAC with built-in policies. To customize the available configuration, you can configure a profile using custom Open Mobile Alliance-Uniform Resource Identifier (OMA-URI) settings. Follow our step-by-step guide to deploy or remove WDAC policies.
(5 mins)
WDAC + Intune + OMA-URI + CSP + Policies + AppLocker
Go deeper on the following MDM-related topics:
Skilling snack: Go cloud first with Windows device management
Skilling snack: Windows Autopilot 101
Skilling snack: Windows Autopatch
Skilling snack: Best practices for shared and frontline Windows devices
Skilling snack: Windows application security
Thanks for putting in your order for this skilling snack on mobile device management! Leave us a comment below or in our skilling snack library with additional topics you’d like to savor. Remember that we add fresh items to the menu every other week. Tell a friend.
Continue the conversation. Find best practices. Bookmark the Windows Tech Community, then follow us @MSWindowsITPro on X and on LinkedIn. Looking for support? Visit Windows on Microsoft Q&A.
Microsoft Tech Community – Latest Blogs –Read More
Latest advancements in Premium SSD v2 and Ultra Azure Managed Disks
We are excited to share the latest advancements in Premium SSD v2 (Pv2) and Ultra disks, the next generation of Azure disk storage. We have enhanced the backup and disaster recovery experience for Pv2 and Ultra disks to help you protect your workloads with ease. In addition, we have improved the security measures for Virtual Machines (VMs) equipped with Pv2 and Ultra.
Pv2 and Ultra disks offer high throughput, high input/output operations per second (IOPS), and low latency for scenarios ranging from input/output (IO)-intensive workloads like SAP High-performance Analytic Appliance (SAP HANA) to general purpose applications across Relational Database Management System (RDBMS), SQL, and NoSQL databases. Pv2 features an unparalleled balance of price and performance—a 23% price-performance improvement over the competing cloud according to a 2023 GigaOm report—and Ultra provides top performance with low sub-millisecond latency.
Check out the full blog here for the latest advancements!
Microsoft Tech Community – Latest Blogs –Read More
are there motor resolver models out there?
Has anyone modeled a dc motor and resolver in multibody or other? Looking for an example rather brute force attempting including learning curve.Has anyone modeled a dc motor and resolver in multibody or other? Looking for an example rather brute force attempting including learning curve. Has anyone modeled a dc motor and resolver in multibody or other? Looking for an example rather brute force attempting including learning curve. resolver MATLAB Answers — New Questions
Don’t understand the reason this code is giving me errors.
The code that is producing the error is below. It says:
>> test
Error: File: test.m Line: 32 Column: 58
Invalid expression. When calling a function or indexing a variable, use parentheses. Otherwise, check for mismatched delimiters.
>> test
Error: File: test.m Line: 30 Column: 60
Invalid expression. When calling a function or indexing a variable, use parentheses. Otherwise, check for mismatched delimiters.
Code that produces errors is below.
clc;
mass = 13.5;
Jx = 0.8244;
Jy = 1.135;
Jz = 1.759;
Jxz = 0.1204;
G = Jx*Jz-Jxz^2;
G1 = Jxz*(Jx – Jy + Jz)/G;
G2 = (Jz*(Jz-Jy)+Jxz^2)/G;
G3 = Jz/G;
G4 = Jxz/G;
G5 = (Jz-Jx)/Jy;
G6 = Jxz/Jy;
G7 = ((Jx-Jy)*Jx +Jxz^2)/G;
G8 = (Jx/G);
p = 0;
q = 0;
r = 0;
l = 0.0000;
m = 0.0000;
n = 0.0000;
tspan = [0 10];
%[t,q] = ode45(@(t,q) (G5*p*r-G6*(p^2-r^2)+m/Jy), tspan,0);
%plot(t,q);
%pdot = (G1*p*q-G2*q*r + G3*l+G4*n);
[t,p]=ode45(@(t,p) (G1*p*q-G2*q*r+G3*l+G4*n),[0 10], 0.1);
plot(t,p,’–r’);
disp(p);
disp(t);
[t,q] = ode45(@(t,r) (G5*p*r-G6*p*p-G6*r*r)+m/Jy), [0 10],0);
plot(t,q,’–g’);
[t,r] = ode45(@(t,r) (G7*p*q-G1*q*r + G4*l+G8*n), [0 10],0);
plot(t,r,’–b’)The code that is producing the error is below. It says:
>> test
Error: File: test.m Line: 32 Column: 58
Invalid expression. When calling a function or indexing a variable, use parentheses. Otherwise, check for mismatched delimiters.
>> test
Error: File: test.m Line: 30 Column: 60
Invalid expression. When calling a function or indexing a variable, use parentheses. Otherwise, check for mismatched delimiters.
Code that produces errors is below.
clc;
mass = 13.5;
Jx = 0.8244;
Jy = 1.135;
Jz = 1.759;
Jxz = 0.1204;
G = Jx*Jz-Jxz^2;
G1 = Jxz*(Jx – Jy + Jz)/G;
G2 = (Jz*(Jz-Jy)+Jxz^2)/G;
G3 = Jz/G;
G4 = Jxz/G;
G5 = (Jz-Jx)/Jy;
G6 = Jxz/Jy;
G7 = ((Jx-Jy)*Jx +Jxz^2)/G;
G8 = (Jx/G);
p = 0;
q = 0;
r = 0;
l = 0.0000;
m = 0.0000;
n = 0.0000;
tspan = [0 10];
%[t,q] = ode45(@(t,q) (G5*p*r-G6*(p^2-r^2)+m/Jy), tspan,0);
%plot(t,q);
%pdot = (G1*p*q-G2*q*r + G3*l+G4*n);
[t,p]=ode45(@(t,p) (G1*p*q-G2*q*r+G3*l+G4*n),[0 10], 0.1);
plot(t,p,’–r’);
disp(p);
disp(t);
[t,q] = ode45(@(t,r) (G5*p*r-G6*p*p-G6*r*r)+m/Jy), [0 10],0);
plot(t,q,’–g’);
[t,r] = ode45(@(t,r) (G7*p*q-G1*q*r + G4*l+G8*n), [0 10],0);
plot(t,r,’–b’) The code that is producing the error is below. It says:
>> test
Error: File: test.m Line: 32 Column: 58
Invalid expression. When calling a function or indexing a variable, use parentheses. Otherwise, check for mismatched delimiters.
>> test
Error: File: test.m Line: 30 Column: 60
Invalid expression. When calling a function or indexing a variable, use parentheses. Otherwise, check for mismatched delimiters.
Code that produces errors is below.
clc;
mass = 13.5;
Jx = 0.8244;
Jy = 1.135;
Jz = 1.759;
Jxz = 0.1204;
G = Jx*Jz-Jxz^2;
G1 = Jxz*(Jx – Jy + Jz)/G;
G2 = (Jz*(Jz-Jy)+Jxz^2)/G;
G3 = Jz/G;
G4 = Jxz/G;
G5 = (Jz-Jx)/Jy;
G6 = Jxz/Jy;
G7 = ((Jx-Jy)*Jx +Jxz^2)/G;
G8 = (Jx/G);
p = 0;
q = 0;
r = 0;
l = 0.0000;
m = 0.0000;
n = 0.0000;
tspan = [0 10];
%[t,q] = ode45(@(t,q) (G5*p*r-G6*(p^2-r^2)+m/Jy), tspan,0);
%plot(t,q);
%pdot = (G1*p*q-G2*q*r + G3*l+G4*n);
[t,p]=ode45(@(t,p) (G1*p*q-G2*q*r+G3*l+G4*n),[0 10], 0.1);
plot(t,p,’–r’);
disp(p);
disp(t);
[t,q] = ode45(@(t,r) (G5*p*r-G6*p*p-G6*r*r)+m/Jy), [0 10],0);
plot(t,q,’–g’);
[t,r] = ode45(@(t,r) (G7*p*q-G1*q*r + G4*l+G8*n), [0 10],0);
plot(t,r,’–b’) error in calling runge kutta 4 function. MATLAB Answers — New Questions
Get help on a function with Copilot in Excel
Greetings, we continue our series of posts this week that show how you can get help from copilot with the chat helper. Copilot in Excel can help you to find the right function for a task and can also help you if you know what function you want to use but need help with an example. We’ll show this by asking Copilot in Excel the following question:
Can you give me an example of how to use =textafter in Excel?
Copilot in Excel analyzes the question and returns the response below, explaining how to use the TEXTAFTER function:
“Certainly! The =TEXTAFTER function in Excel is used to extract text that comes after a specific delimiter in a given string. Here’s an example of how to use it:
Suppose you have a list of email addresses in column A, and you want to extract the domain part of the email (the part after the “@”). You can use the =TEXTAFTER function to do this.
Here’s the formula you would use in cell B1 and then drag down to apply to other cells in the column:
=TEXTAFTER(A1, “@”, , 1)
A1 is the cell containing the email address.
“@” is the delimiter after which you want to extract the text.
The third argument is left blank to use the default behavior, which is to return the text after the first occurrence of the delimiter.
1 specifies that you want to extract the text after the first occurrence of the “@” symbol.
If the “@” symbol is not found in the text, the function will return an error by default. You can modify the behavior by adding an additional argument to specify a different outcome when the delimiter is not found.”
Over the coming weeks I will be sharing more examples of what you can do with Copilot in Excel.
Thanks for reading,
Microsoft Excel Team
*Disclaimer: If you try these types of prompts and they do not work as expected, it is most likely due to our gradual feature rollout process. Please try again in a few weeks.
Greetings, we continue our series of posts this week that show how you can get help from copilot with the chat helper. Copilot in Excel can help you to find the right function for a task and can also help you if you know what function you want to use but need help with an example. We’ll show this by asking Copilot in Excel the following question:
Can you give me an example of how to use =textafter in Excel?
Copilot in Excel analyzes the question and returns the response below, explaining how to use the TEXTAFTER function:
Picture of Copilot in Excel pane showing the above prompt with a description of the TEXTAFTER function and how to use it.
“Certainly! The =TEXTAFTER function in Excel is used to extract text that comes after a specific delimiter in a given string. Here’s an example of how to use it:
Suppose you have a list of email addresses in column A, and you want to extract the domain part of the email (the part after the “@”). You can use the =TEXTAFTER function to do this.
Here’s the formula you would use in cell B1 and then drag down to apply to other cells in the column:
=TEXTAFTER(A1, “@”, , 1)
A1 is the cell containing the email address.
“@” is the delimiter after which you want to extract the text.
The third argument is left blank to use the default behavior, which is to return the text after the first occurrence of the delimiter.
1 specifies that you want to extract the text after the first occurrence of the “@” symbol.
If the “@” symbol is not found in the text, the function will return an error by default. You can modify the behavior by adding an additional argument to specify a different outcome when the delimiter is not found.”
Over the coming weeks I will be sharing more examples of what you can do with Copilot in Excel.
Thanks for reading,
Microsoft Excel Team
*Disclaimer: If you try these types of prompts and they do not work as expected, it is most likely due to our gradual feature rollout process. Please try again in a few weeks. Read More
Windows Boot Manager not updated
KB5025885 outlines how to update Windows Boot Manager, but build 26257 still has a Windows Boot Manager that is signed with the old “PCA 2011” certificate.
The manual processes in KB5025885 are a real pain (and don’t scale) so it would be very annoying if this is not fixed before RTM.
KB5025885 outlines how to update Windows Boot Manager, but build 26257 still has a Windows Boot Manager that is signed with the old “PCA 2011” certificate. The manual processes in KB5025885 are a real pain (and don’t scale) so it would be very annoying if this is not fixed before RTM. https://support.microsoft.com/en-us/topic/kb5025885-how-to-manage-the-windows-boot-manager-revocations-for-secure-boot-changes-associated-with-cve-2023-24932-41a975df-beb2-40c1-99a3-b3ff139f832d Read More
Vlookup for excel continuously add15 sheet with macro
Sub santo()
Set ws = Worksheets(“Filter”)
LastRow = Cells.Find(“*”, , xlValues, , xlRows, xlPrevious).Row
For i = 1 To 15
a = Array(ws.Range(a(i) & “1”).Value, ws.Range(a(i) & “2”).Value, ws.Range(a(i) & “3”).Value, ws.Range(a(i) & “4”).Value)
Sheets.Add.Name = a(0)
With Worksheets(“Sales Data”).Range(“A1” & LastRow)
.AutoFilter Field:=4, Criteria1:=Array(a(1), a(2), a(3)), Operator:=xlFilterValues
.SpecialCells(xlCellTypeVisible).Copy Destination:=Sheets(a(0)).Range(“A1”)
End With
Next
End Sub
Sub santo()
Set ws = Worksheets(“Filter”)
LastRow = Cells.Find(“*”, , xlValues, , xlRows, xlPrevious).Row
For i = 1 To 15
a = Array(ws.Range(a(i) & “1”).Value, ws.Range(a(i) & “2”).Value, ws.Range(a(i) & “3”).Value, ws.Range(a(i) & “4”).Value)
Sheets.Add.Name = a(0)
With Worksheets(“Sales Data”).Range(“A1” & LastRow)
.AutoFilter Field:=4, Criteria1:=Array(a(1), a(2), a(3)), Operator:=xlFilterValues
.SpecialCells(xlCellTypeVisible).Copy Destination:=Sheets(a(0)).Range(“A1”)
End With
Next
End Sub Read More
Sharepoint File collaboration report
Is there a way to find out what documents in SharePoint (org wide) that have had multiple users editing them and or multiple users viewing them at one time. OR is there anything that can be found about all users that have edited a document without having to manually touch each document.
Is there a way to find out what documents in SharePoint (org wide) that have had multiple users editing them and or multiple users viewing them at one time. OR is there anything that can be found about all users that have edited a document without having to manually touch each document. Read More
Invalid requesting-user-name value after system reboot (IPP Everywhere)
When I add a IPP Printer, it works fine until I restart OS, after system reboot the requesting-user-name contains WORKGROUPVM132786998$ so it’s domaincomputerName$. And the user cannot be correctly recognized.
Steps to reproduce it:
1. Add IPP printer
2. Print Test page
3. Restart system
4. Print Test page
Current result if inspecting by Wireshark:
at point 2. requesting-user-name contains correct “Administrator” value
at point 4. requesting-user-name contains incorrect “workgroupVM132786998$
Expected result: in both cases correct value Administrator must be present.
Currently only restart print spooler or recreating printer solves the issue. (See attached wireshark captures before and after print spooler restart)
Note: The issue is not reproducible in build 22H2 22621.3447. I’m able to reproduce it in 23H2 22635.4076.
Same error is observed here (comment from Douglas Kosovic): https://learn.microsoft.com/en-us/answers/questions/1428508/ipp-everywhere-ipp-2-0-server-basic-authentication
Attached screenshots from wireshark captures when it’s happening and after restarting Print Spooler.
When I add a IPP Printer, it works fine until I restart OS, after system reboot the requesting-user-name contains WORKGROUPVM132786998$ so it’s domaincomputerName$. And the user cannot be correctly recognized. Steps to reproduce it:1. Add IPP printer2. Print Test page3. Restart system4. Print Test page Current result if inspecting by Wireshark:at point 2. requesting-user-name contains correct “Administrator” valueat point 4. requesting-user-name contains incorrect “workgroupVM132786998$ Expected result: in both cases correct value Administrator must be present.Currently only restart print spooler or recreating printer solves the issue. (See attached wireshark captures before and after print spooler restart) Note: The issue is not reproducible in build 22H2 22621.3447. I’m able to reproduce it in 23H2 22635.4076. Same error is observed here (comment from Douglas Kosovic): https://learn.microsoft.com/en-us/answers/questions/1428508/ipp-everywhere-ipp-2-0-server-basic-authenticationAttached screenshots from wireshark captures when it’s happening and after restarting Print Spooler. Read More
What does Autopilot add to Intune
After some testing I do not see the advantage/benefits of deploying Autopilot.
To me it seems that if a user logs in on a new, unconfigured device with the Entra account we give them, configured in Intune, we end up with the same configuration as with using Autopilot, except of course for the things configured in the section Autopilot, so Deployment Profiles, Enrollment Status Page, etc.
I must be missing something?
After some testing I do not see the advantage/benefits of deploying Autopilot.To me it seems that if a user logs in on a new, unconfigured device with the Entra account we give them, configured in Intune, we end up with the same configuration as with using Autopilot, except of course for the things configured in the section Autopilot, so Deployment Profiles, Enrollment Status Page, etc.I must be missing something? Read More
Error message when uploading image to Copilot
I cannot upload images. It kept giving me this message “The file couldn’t be uploaded. Please try again” How can I solve this?
I cannot upload images. It kept giving me this message “The file couldn’t be uploaded. Please try again” How can I solve this? Read More
New GPU-enabled Windows 365 experiences
New Windows 365 GPU-enabled Cloud PCs can help maximize productivity for anyone in graphics-intensive fields such as graphic design, video editing, 3D modeling, and data visualization. With Windows 365, you can stream your Windows experience from the Microsoft cloud to any device and benefit security, increase connection speeds, and reduced latency.
Run demanding workloads with GPU-enabled Cloud PCs.
Benefit from on-device generative AI experiences and machine learning models — like the Small Language Model, PHI 3 Mini running offline and generating responses based on open world training. Take a look.
Benefit from fast data transfer.
Minimal latency on Cloud PCs with Windows 365, connected to the Microsoft Cloud’s high-speed network. See it here.
Keep data separate from interacting devices.
Enable secure integrations like local printing, file transfer, webcam support, and clipboard access. Check out Windows 365 GPU-enabled Cloud PCs.
Watch our video here:
QUICK LINKS:
00:00 — Cloud PC in Windows 365 can be GPU-enabled
00:28 — Run compute-intensive apps
00:50 — Run demanding workloads
01:37 — Fast data transfer with reduced latency
02:19 — Windows 365 GPU options
03:30 — Shared utilization
03:50 — Data remains separate
04:13 — Windows 365 new watermark support
04:28 — Windows 365 management
05:01 — Wrap up
Link References
Get started at https://windows365.com
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:
-Can a Cloud PC in Windows 365 securely run your most demanding workloads? Well, if you’re new to Windows 365, it lets you securely stream your Windows experience and access your personalized apps, your content, and your settings to literally any device from the Microsoft Cloud.
-And now, your Cloud PC in Windows 365 can be GPU-enabled using some of the most powerful accelerators in the cloud today. Here, you’re seeing an older generation MacBook Air, and using our Cloud PC, we can bypass any device compute limitations to deliver a GPU-powered Windows 365 experience to run your most compute-intensive graphic design applications, high-end video editing and rendering, and even sophisticated 3D modeling.
-In fact, Cloud PCs can now run your most demanding workloads. Here in our GPU-enabled Cloud PC, you’re seeing an industrial digital twin example, which would typically have needed to run on a powerful physical PC, and notice the textures, shadows and materials all rendered in real time.
-And when you select our most powerful GPU option, you can also take advantage of some of the latest on-device generative AI experiences and machine learning models like you’re seeing here with the Small Language Model PHI 3 Mini running in offline mode and generating responses based on its open world training. And this opens up new possibilities for data scientists and developers. Additionally, data transfer is fast with reduced latency because Cloud PCs are connected to the blazing fast network backbone of the Microsoft Cloud.
-Here, we’re running the Speedtest app in our Cloud PC, and in this case, both download and upload speeds vastly exceed the average broadband speeds on your home or office network. So no matter your local network limitations, you benefit from minimal latency and speed, making them ideally suited for large file transfers. And each GPU-enabled Cloud PC by default is configured with a stateless ephemeral disc for local in-session caching for lower latency with faster read and write speeds.
-Now, there are a few Windows 365 GPUs to choose from. Options include Standard, Super and Max with some key differences. Standard is comparable to a PC with entry-level discrete graphics. It uses a shared GPU and is optimized for basic graphic acceleration, supporting a 4K display or two 1080p displays. It includes a minimum four vCPUs, 16 gigabytes of RAM, eight gigabytes of video RAM and 512 gigabytes of storage.
-Then Super elevates the performance to something more comparable to a PC with a mid-tier graphics card. And it also uses a shared GPU, supporting up to four 4K connected displays, and it has at least eight vCPUs, 56 gigabytes of RAM, 12 gigabytes of video RAM and a terabyte of storage. Then, Max takes it up to the next level, providing a fully dedicated GPU for the highest demands.
-It boosts the specs to a minimum of 16 vCPUs, 110 gigabytes of RAM, 16 gigabytes of video RAM and a terabyte of storage. Additionally, new Windows 365 options are available for shared utilization scenarios to use a GPU-enabled Cloud PC. In this case, a GPU-enabled Cloud PC can be shared with up to three users who can take turns in accessing it one by one as they need to, helping to optimize costs while keeping resources utilized.
-Importantly, your data remains separate from the devices interacting with your Cloud PCs, and device integrations like printing to local printers, file transfer, support for webcam and microphone pass through, as well as copy paste to and from the local clipboard can be individually enabled by admin policies and controlled by users.
-Additionally, Windows 365’s new watermark support helps protect your visual assets from people taking photos of screens or using display capture cards or screen recording software with a QR code that points back to the source Cloud PC. For device administrators, Windows 365 management is integrated in Microsoft Intune and this lets you manage them without requiring traditional VDI skills in the same place you would manage your physical devices.
-From Intune, you can centrally monitor your entire Cloud PC fleet, and, of course, you have complete policy configuration control over security and compliance settings in your managed Cloud PCs. The Intune admin center lets you view key indicators and trends for end user experiences and easily diagnose and troubleshoot any issues.
-So, that’s how Windows 365 GPU-powered Cloud PCs can securely run your most demanding workloads by leveraging some of the most powerful accelerators in the cloud today on any device. To learn more and get started, check out Windows365.com and thank you for watching.
Microsoft Tech Community – Latest Blogs –Read More
Empowering.Cloud Community Update – September 2024
Check out the latest expert briefings, Teams Insider Podcast, Microsoft Teams Operator Connect updates and upcoming community events.
Expert Briefings
Searching for Teams Messages
https://app.empowering.cloud/briefings/372/searching-for-teams-messages
In our latest community briefing, MVP Nikki Chapple explains the three methods to search Microsoft Teams messages.
The three methods to search for Teams Messages are:
User search, content search and eDiscovery Standard
User Search: For end-users, quick and shows threaded conversation, security trimmed
Content Search and eDiscovery Standard: For admins, searches entire tenant but is limited to individual messages
eDiscovery Premium: Comprehensive search for admins, shows threaded conversations, detailed export
Microsoft Teams Monthly Update August 2024
https://app.empowering.cloud/briefings/373/Microsoft-Teams-Monthly-Update-August-2024
MVP Tom Arbuthnot runs through the latest Microsoft news and updates you need to know this month.
Microsoft Earnings FY24 Q4 Modern Work Insights
Multiple Teams Client Improvements
Custom Emojis and Reactions
Copilot Dashboard for all M365 Copilot users
Speaker recognition for all Teams Rooms on Windows
Microsoft Teams Rooms on Windows 5.1.24.0
Microsoft Teams Devices updates
Android Teams Room and Teams Phone support date changes
Teams Displays end of support September 2025
Roadmap Watch
Microsoft Teams Insider Podcast
Microsoft Mesh, Real-world Use cases and AI Futures
Jeremy Dalton, Head of Immersive Technologies at PWC and Cathy Moya, Senior Customer Success Program Manager – Mesh Team at Microsoft, dive into Microsoft Mesh adoption and use cases and how virtual reality and virtual worlds can enhance collaboration and communication.
AV and UC Converging with John Bailey, Senior Vice President AVI-SPL
John Bailey, Senior Vice President of Technology and Innovation at AVI-SPL, discusses industry changes and the convergence of AV and IT.
Microsoft Devices Ecosystem Platform (MDEP) Explained
Yoav Barzilay, Microsoft Senior Product Manager, explains the Microsoft Devices Ecosystem Platform (MDEP) and how it fits in the Microsoft Teams devices ecosystem.
Lessons from BT’s Large Scale Microsoft Teams Rooms Deployment
James Robinson, Director & Principal AV Consultant at AVenture Audio Visual, shares his extensive previous experiences with deploying Microsoft Teams Rooms at scale at BT.
Microsoft Teams Operator Connect Updates
The race to 100 countries or 100 operators continues with both countries and operators at the 97 mark. There’s not too much to report here this month, although providers continue to be extending their Microsoft Teams Operator Connect coverage to more countries.
Providers Change:
Momentum – Canada and USA
Country Changes:
HALO – USA
LoopUp – Republic of Korea
Orange Business Services – Mexico
Tata Communicaitons – Uganda, South Africa and Dominican Republic
Check out our full Power BI report of all the Operators here:
https://members.empowering.cloud/microsoft-teams-operator-connect-providers
Upcoming Community Events
Teams Fireside Chat with Ron Pessner – 12th September | 16:00 BST | Virtual
Join other community members for the September Teams Fireside Chat where Ron Pessner, Microsoft Corporate Vice President – Product Management, will be joining us to share his insights.
Teams Fireside Chat occurs on the 2nd Thursday of every month at 16:00 UK. For each session, we are joined by a guest speaker from Microsoft who shares their insights. In the second half, we open the floor to discussion and Q&A, giving you the chance to get your questions answered by the experts.
Microsoft Teams Devices Ask Me Anything
EMEA/NA – 16th September | 16:00 BST | Virtual
Hosts: MVP Graham Walsh, Michael Tressler, Jimmy Vaughan
APAC – 17th September| 15:30 AEST | Virtual
Hosts: MVP Graham Walsh, Phil Clapham, Andrew Higgs, Justin O’Meara
Microsoft Teams Devices Ask Me Anything is a monthly community which gives you all an update on the important and Microsoft Teams devices news, as well as the chance to ask questions and get them answered by the experts. We have 2 sessions to cover different time zones, so there’s really no excuse not to come along to at least one!
Microsoft Tech Community – Latest Blogs –Read More
Discover the New Multi-Lingual, High-Quality Phi-3.5 SLMs
The Phi-3 model collection is the latest in Microsoft’s family of Small Language Models (SLMs). They are designed to be highly capable and cost-effective, outperforming models of similar and larger sizes across various benchmarks in language, reasoning, coding, and math. The availability of Phi-3 models expands the selection of high-quality models for Azure customers, offering more practical choices as they compose and build generative AI applications. Since the launch in April 2024, we have received lots of valuable feedback from our customers and community members on areas for improvement in the Phi-3 models. Today, we are proud to announce Phi-3.5-mini, Phi-3.5-vision, and a new member to the Phi family, Phi-3.5-MoE, a Mixture-of-Experts (MoE) model. Phi-3.5-mini enhances multi-lingual support with a 128K context length. Phi-3.5-vision improves multi-frame image understanding and reasoning, boosting performance on single-image benchmarks. Phi-3.5-MoE, featuring 16 experts and 6.6B active parameters, provides high performance, reduced latency, multi-lingual support, and robust safety measures, excelling over larger models while upholding Phi models efficacy.
Phi-3.5-MoE: Mixture-of-Experts
Phi-3.5-MoE is the latest addition to the Phi model family. It comprises 16 experts, each containing 3.8B parameters. With a total model size of 42B parameters, it activates 6.6B parameters when using two experts. This MoE model outperforms a similarly sized dense model in terms of quality and performance. It supports over 20 languages. Like its Phi-3 counterparts, the MoE model employs a robust safety post-training strategy, using a mix of open-source and proprietary synthetic instruction and preference datasets. This post-training process combines Supervised Fine-Tuning (SFT) and Direct Preference Optimization (DPO), utilizing both human-labeled and synthetic datasets. These include datasets focused on helpfulness and harmlessness, as well as multiple safety categories. Phi-3.5-MoE also supports a context length of up to 128K, enabling it to handle numerous long-context tasks.
To understand the model quality, we compare Phi-3.5-MoE with a set of models over a variety of benchmarks as shown in Table 1:
We take a closer look at different categories public benchmarks datasets at the table below:
Phi-3.5-MoE with only 6.6B active parameters achieves a similar level of language understanding and math as much larger models. Moreover, the model outperforms bigger models in reasoning capability. The model provides good capacity for finetuning for various tasks. Table 3 highlights multi-lingual capability of the Phi-3.5-MoE on multi-lingual MMLU, MEGA, and multi-lingual MMLU-pro datasets. Overall, we observed that even with just 6.6B active parameters, the model is very competitive on multi-lingual tasks in comparison to other models with much bigger active parameters.
Multi-lingual Capability
The table below shows multi-lingual MMLU scores in some of the supported languages.
Phi-3.5-mini
Phi-3.5-mini model has undergone further pre-training using multi-lingual synthetic and high-quality filtered data. This was followed by a series of post-training steps which included Supervised Fine-Tuning (SFT), Proximal Policy Optimization (PPO), and Direct Preference Optimization (DPO). These processes utilized a combination of human-labeled, synthetic, and translated datasets.
Model Quality
When diving into the capabilities of language models, it’s crucial to understand how they stack up against one another. That’s why we put Phi-3.5-mini to the test alongside a selection of recent top performing larger models, utilizing our internal benchmark platform. In a high-level overview, Table 1 provides a snapshot of the model quality on key benchmarks. Despite its compact size of just 3.8B parameters, this efficient model not only matches but often surpasses the performance of other models of larger sizes.
Multi-lingual Capability
Phi-3.5-mini is our latest 3.8B model update. The model used additional continual pre-training and post-training data leading to substantial gains on multi-lingual, multi-turn conversation quality, and reasoning capability. The model has been trained on selective set of languages listed here: Arabic, Chinese, Czech, Danish, Dutch, English, Finnish, French, German, Hebrew, Hungarian, Italian, Japanese, Korean, Norwegian, Polish, Portuguese, Russian, Spanish, Swedish, Thai, Turkish and Ukrainian.
Table 6 below highlights multi-lingual capability of Phi-3.5-mini on the average language specific scores of multi-lingual MMLU, MGSM, MEGA, and multi-lingual MMLU-pro datasets.
Table 7 below shows Multi-lingual MMLU scores for some of the supported languages.
Phi-3.5-mini shows significant improvement over Phi-3-mini on multi-lingual support. Arabic, Dutch, Finnish, Polish, Thai and Ukrainian received the most boost from the new Phi version with 25-50% improvement in performance. Putting these into wider perspective, Phi-3.5-mini shows top performance for any sub-8B model, in English as well as several languages. It is worth noting that the model uses 32K vocabulary and optimized for higher resource languages above, it is not recommended to be used for low resource languages without further fine-tuning.
Long Context
Phi-3.5-mini, with a 128K context length support, excels in tasks like summarizing long documents or meeting transcripts, long document-based QA, and information retrieval. Phi-3.5 performs better than the Gemma-2 family, which supports only an 8K context length. Additionally, Phi-3.5-mini is highly competitive with much larger open-weight models such as Llama-3.1-8B-instruct, Mistral-7B-instruct-v0.3, and Mistral-Nemo-12B-instruct-2407. Table 8 lists various long-context benchmarks.
Ruler: a retrieval-based benchmark for long context understanding
RepoQA: a benchmark for long context code understanding
With only 3.8B parameters, 128K context length, and multi-lingual support, Phi-3.5-mini-instruct is the only model in this category. It is worth noting that we opted to support more languages while maintaining English performance on various tasks. Due to limited model capacity, this may lead to English knowledge on the model being better than other languages. For multi-lingual knowledge intense tasks, we recommend using the model in RAG setup.
Phi-3.5-vision with Multi-frame Input
Phi-3.5-vision introduces cutting-edge capabilities for multi-frame image understanding and reasoning, developed based on invaluable customer feedback. This innovation empowers detailed image comparison, multi-image summarization/storytelling, and video summarization, offering a wide array of applications across various scenarios.
For example, see the model output for summarization of multiple slides:
Remarkably, Phi-3.5-vision has demonstrated significant performance improvements in numerous single-image benchmarks. For example, it boosted the MMMU performance from 40.4 to 43.0 and improved the MMBench performance from 80.5 to 81.9. Additionally, the document understanding benchmark TextVQA saw an increase from 70.9 to 72.0.
The following tables illustrate the detailed comparison results on two renowned multi-image/video benchmarks, showcasing the enhanced performance metrics. It is worth noting that Phi-3.5-Vision is not optimized for multi-lingual use cases. It is advised not to use it for multi-lingual scenarios without further fine-tuning.
Safety
The Phi-3 family of models were developed in accordance with the Microsoft Responsible AI Standard, which is a company-wide set of requirements based on the following six principles: accountability, transparency, fairness, reliability and safety, privacy and security, and inclusiveness. Like the previous Phi-3 models, a multi-faceted safety evaluation and safety post-training approach was adopted, with additional measures taken to account for multi-lingual capabilities of this release. Our approach to safety training and evaluations including testing across multiple languages and risk categories is outlined in the Phi-3 Safety Post-Training Paper. While the Phi-3 models benefit from this approach, developers should apply responsible AI best practices, including mapping, measuring, and mitigating risks associated with their specific use case and cultural and linguistic context.
Optimized Variants
ONNX Runtime provides optimized inference for the Phi family of models. You can optimize Phi-3.5-mini on various hardware targets today using this example. Stay tuned for updated ONNX variants of latest Phi-3.5 models in the coming weeks.
More Predictable Outputs
We are bringing Guidance to the Phi-3.5-mini serverless endpoint offering in Azure AI Studio to make outputs more predictable through defining the structure tailored to an application. With Guidance, you can eliminate expensive retries, and can, for example, constrain the model to select from pre-defined lists (e.g., medical codes), restrict outputs to direct quotes from provided context, or follow in any regex. Guidance steers the model token by token in the inference stack, reducing cost and latency by 30-50%, which makes it a unique and valuable add-on to the Phi-3-mini serverless endpoint.
Closing Remarks
Phi-3.5-mini has emerged as a unique offering in the LLMs landscape, boasting only 3.8B parameters, a substantial 128K context length, and multi-lingual support. Phi-3.5-mini represents a milestone in creating efficient, multi-lingual models, striking a delicate balance between broad language support and focused performance in English. Given the small model capacity, users may observe that the density of English knowledge in the model surpasses that of other languages. When approaching multi-lingual, knowledge-intensive tasks, it’s advisable to utilize Phi-3.5-mini within a Retrieval-Augmented Generation (RAG) setup. This configuration can significantly enhance the model’s performance across different languages by leveraging external data sources, thereby mitigating the language-specific limitations imposed by its compact architecture.
Phi-3.5-MoE, featuring 16 small experts, delivers high-quality performance and reduced latency, supports 128k context length, and multiple languages with strong safety measures. It surpasses larger models and can be customized for various applications through fine-tuning, all while maintaining efficiency with 6.6B active parameters.
Phi-3.5-vision introduces advancements in multi-frame image understanding and reasoning, enhancing single-image benchmark performance.
The Phi-3.5 model family provides cost-effective, high-capability options for the open-source community and Azure customers, pushing the boundaries of small language models and generative AI.
Microsoft Tech Community – Latest Blogs –Read More