- 124
- 60 004
Brett Gillett
Canada
เข้าร่วมเมื่อ 11 ธ.ค. 2019
Building a Custom Knowledge Base with Amazon Bedrock - Part III
Welcome back to another CloudFormation and AWS Bedrock project session! After a brief hiatus, we dive back into our project to fix some permissions issues and enhance our custom knowledge base.
This episode covers: resolving permissions for Bedrock deployments, configuring OpenSearch serverless collections, manually creating indexes, setting up server roles, and testing Bedrock's capabilities using custom data sources. Follow along to see the troubleshooting process, solutions, and a successful deployment.
OrbitWEEKLY - AWS News & Learning Opportunities
► curiousorbit.com/orbitweekly/
PART 1
► th-cam.com/video/smFJ5v9v9rI/w-d-xo.html
PART 2
► th-cam.com/video/afWP64LbhU8/w-d-xo.html
00:00 Introduction and Stream Recap
00:31 Fixing Previous Issues
01:06 OpenSearch Serverless Configuration
01:49 Creating the Knowledge Base
03:41 Switching to NeoVim
04:13 Updating Policies and Running Pipelines
10:33 Manual Index Creation
17:09 Troubleshooting Permissions
28:03 Configuring Vector Fields
42:12 Final Adjustments and Testing
44:38 Troubleshooting the Missing S3 Bucket
46:30 Configuring the S3 Data Source
47:06 Setting Up Data Deletion Policies
48:14 Defining Data Source Configuration
55:37 Handling Permissions and Errors
59:59 Uploading and Syncing Data
01:06:16 Final Adjustments and Testing
01:23:19 Querying the Knowledge Base
01:26:04 Conclusion and Future Plans
This episode covers: resolving permissions for Bedrock deployments, configuring OpenSearch serverless collections, manually creating indexes, setting up server roles, and testing Bedrock's capabilities using custom data sources. Follow along to see the troubleshooting process, solutions, and a successful deployment.
OrbitWEEKLY - AWS News & Learning Opportunities
► curiousorbit.com/orbitweekly/
PART 1
► th-cam.com/video/smFJ5v9v9rI/w-d-xo.html
PART 2
► th-cam.com/video/afWP64LbhU8/w-d-xo.html
00:00 Introduction and Stream Recap
00:31 Fixing Previous Issues
01:06 OpenSearch Serverless Configuration
01:49 Creating the Knowledge Base
03:41 Switching to NeoVim
04:13 Updating Policies and Running Pipelines
10:33 Manual Index Creation
17:09 Troubleshooting Permissions
28:03 Configuring Vector Fields
42:12 Final Adjustments and Testing
44:38 Troubleshooting the Missing S3 Bucket
46:30 Configuring the S3 Data Source
47:06 Setting Up Data Deletion Policies
48:14 Defining Data Source Configuration
55:37 Handling Permissions and Errors
59:59 Uploading and Syncing Data
01:06:16 Final Adjustments and Testing
01:23:19 Querying the Knowledge Base
01:26:04 Conclusion and Future Plans
มุมมอง: 1 054
วีดีโอ
The Impact of Gen AI on Weather Forecasting, Flood Predictions and AWS Updates
มุมมอง 514 หลายเดือนก่อน
In this week's episode, Brett and Travers discuss the latest technology news and notable AWS updates and examine the societal impact of generative AI on various fields. Key AWS updates include command line auto-completions in Amazon QCLI, extended API gateway integration timeout limits, and new features in health imaging and Bedrock's Titan Text model. The episode also explores Google's researc...
CloudFormation, Amazon Bedrock and GenAI
มุมมอง 715 หลายเดือนก่อน
In this episode of Talking Cloud, Brett and Travers dive into the latest AWS updates and GenAI news, discussing CloudFormation improvements, Bedrock’s new model, CloudTrail tips, and the future of AI-enhanced Alexa. They also explore the potential of AI replacing CEOs, the growing AI chip market, and intriguing insights into AI model understanding and scaling. This episode is packed with practi...
Introduction to Generative AI on AWS
มุมมอง 3705 หลายเดือนก่อน
In this informative webinar, Brett introduces the fundamentals of Generative AI (Gen AI) on AWS. Brett outlines the basic terminology and concepts necessary to understand Gen AI. He explains that Gen AI is a subset of AI designed to create new content, such as text, images, music, and code, using pre-trained large foundational models. The session highlights AWS’s Bedrock and Amazon Q as essenti...
ChatGPT Voice, NVIDIA Earnings, and Microsoft Recall
มุมมอง 685 หลายเดือนก่อน
In this episode of Talking Cloud, host Brett shares the latest updates from AWS, including improvements to Control Tower, new features in Bedrock, and updates in the Simple Email Service (SES) and Amazon Security Lake. He also discusses general AI news, such as the controversy over ChatGPT's new voice, NVIDIA's financial success with their GPUs, and a new 'Recall' feature in Windows 11 that rec...
AWS Updates, GPT-4 Omni Demos, and Avoiding Suprise Cloud Bills
มุมมอง 625 หลายเดือนก่อน
In this special long weekend edition of Talking Cloud, Brett and Traver bring you the latest and most significant updates from AWS and AI. The episode starts with discussions on the recent changes to Amazon S3 error codes, Amplify CLI security fixes and enhancements in VPC Flow Logs for ECS. They highlight the importance of setting up AWS budgets and anomaly detection to avoid unexpected cloud ...
Protecting Your AMIs from Accidental Deregistration
มุมมอง 345 หลายเดือนก่อน
In this video, learn how to protect your Amazon Machine Images (AMIs) from accidental deregistration. We'll walk you step-by-step through enabling deregistration protection through the AWS Management Console and programmatically using AWS CLI. By following these simple instructions, you can add an extra layer of security to your AWS account and prevent unintended deletions.
Building a Custom Knowledge Base with Amazon Bedrock - Part II
มุมมอง 5905 หลายเดือนก่อน
In this episode, Brett continues to work on building an Amazon Bedrock knowledge base. He starts by addressing previous issues with the OpenSearch vector database, explaining the steps taken to correct deployment regions and resolve naming conflicts in CloudFormation stacks. Brett also emphasizes deploying OpenSearch collections within a VPC for enhanced security. As the episode progresses, Bre...
Gen AI Drives Cloud Adoption, Unlocks Biotech Breakthroughs & Transforms Businessvideo
มุมมอง 3475 หลายเดือนก่อน
In this episode of Talking Cloud, hosts dive into various topics surrounding GenAI and cloud technology. They start with a movie review segment, discussing 'Unfrosted,' a Netflix movie about the invention of the Pop Tart, featuring Jerry Seinfeld and Hugh Grant. The conversation shifts to Star Wars marathons and the Marvel universe before transitioning into AWS news, discussing the new Amazon T...
How to Remove Public IP Addresses from AWS EC2 Instances
มุมมอง 1.5K5 หลายเดือนก่อน
In this video, learn how to remove public IP addresses from EC2 instances with just a few clicks. This update is important for two reasons - improving security by reducing ingress points and lowering costs since AWS now charges for public IPs. You'll see a step-by-step demo of removing a public IP address from an existing EC2 instance using the AWS Management Console and the AWS Command Line In...
AWS Q Services, Research Paper Concerns, and Doctor's AI Assistant
มุมมอง 745 หลายเดือนก่อน
On this episode of Talking Cloud, Brett and Travers dive deep into the latest Gen AI news and updates from AWS. They discuss the new AWS Q services like Q Business, Q Developer, and Q Apps, which reimagine AI for business, developers, and application creation using natural language. They examine concerns around AI's use in academic research papers, with some studies finding that up to 17% of pa...
Building a Custom Knowledge Base with Amazon Bedrock - Part 1
มุมมอง 2.7K6 หลายเดือนก่อน
🔍 Unlock the power of custom knowledge bases with Amazon Bedrock! This exciting tutorial series will walk you through building a tailored knowledge base from scratch. 📚 Part 1 kicks off with a deep dive into the prerequisites and setup. We'll show you how to: - Gather and prepare your source documents (like blog posts in markdown format) 📑 - Set up a vector store using Amazon OpenSearch Serverl...
Exploring GenAI: AWS Services, Skills and Getting Started
มุมมอง 926 หลายเดือนก่อน
In this special edition of the Talking Cloud podcast, hosts Brett and Travers are joined by AWS Senior Partner Solution Architects Deborshi Choudhury and Gerardo Vasquez to dive deep into the world of Generative AI (GenAI). The episode explores the differences between traditional AI and GenAI, real-world examples of GenAI in action, and how organizations can leverage this technology to improve ...
Claude vs. ChatGPT, Eclipse Impact, and Atlas Rising
มุมมอง 396 หลายเดือนก่อน
Brett & Travers discuss their latest entertainment picks, including the series 'Fallout' and movies such as 'Footloose' and 'Day Shift'. The episode pivots to a deep dive into recent advancements and applications of AI and machine learning, specifically highlighting the potential of AI in drug design for diseases like Parkinson's, achieving a significant reduction in cost and development time. ...
AI Chip Wars, Solo D&D, and Fallout
มุมมอง 626 หลายเดือนก่อน
This week, Brett and Travers start off talking about the new Fallout TV series on Amazon Prime, with Brett sharing his positive impressions of the first episode. They then cover several AWS news updates, including the ability to tag CloudWatch alarms via CloudFormation. Deploying Bedrock knowledge bases through CloudFormation, and new top-level domains supported by Route 53 like .beer and .fun....
Bedrock, CloudWatch and Lambda updates and potential impacts of AI
มุมมอง 476 หลายเดือนก่อน
Bedrock, CloudWatch and Lambda updates and potential impacts of AI
The Impact and Future of GenAI in Enterprises: Insights and Challenges
มุมมอง 627 หลายเดือนก่อน
The Impact and Future of GenAI in Enterprises: Insights and Challenges
The Accelerating AI Race: AWS's Ambitious 99% Plan and Lightning-Fast Breakthroughs
มุมมอง 267 หลายเดือนก่อน
The Accelerating AI Race: AWS's Ambitious 99% Plan and Lightning-Fast Breakthroughs
AWS CloudFormation, AI Model Developments, and Nicolas Cage - Talking Cloud EP12
มุมมอง 367 หลายเดือนก่อน
AWS CloudFormation, AI Model Developments, and Nicolas Cage - Talking Cloud EP12
Talking Cloud - Chatbots, Autonomous Agents, and Humanoid Robots
มุมมอง 287 หลายเดือนก่อน
Talking Cloud - Chatbots, Autonomous Agents, and Humanoid Robots
Amazon Q GenAI assistant - added data connectors and more scripting.
มุมมอง 8118 หลายเดือนก่อน
Amazon Q GenAI assistant - added data connectors and more scripting.
GenAI, Gartner, CodeWhisperer, Chatbots and AWS SRA
มุมมอง 528 หลายเดือนก่อน
GenAI, Gartner, CodeWhisperer, Chatbots and AWS SRA
Building and Testing the AmazonQ AI Assistant for Business
มุมมอง 1.1K8 หลายเดือนก่อน
Building and Testing the AmazonQ AI Assistant for Business
OrbitLEARN: Intro to AWS Identity and Access Management (IAM)
มุมมอง 978 หลายเดือนก่อน
OrbitLEARN: Intro to AWS Identity and Access Management (IAM)
Deep Dive into Gen AI, AWS Updates and Emerging Abilities of Large Language Models
มุมมอง 718 หลายเดือนก่อน
Deep Dive into Gen AI, AWS Updates and Emerging Abilities of Large Language Models
Talking Cloud: Debating Infrastructure as Code, Exploring AWS News & Updates
มุมมอง 168 หลายเดือนก่อน
Talking Cloud: Debating Infrastructure as Code, Exploring AWS News & Updates
Mastering Amazon Q Business Edition: From Application Creation to AI Assistant - Session 1
มุมมอง 1.3K8 หลายเดือนก่อน
Mastering Amazon Q Business Edition: From Application Creation to AI Assistant - Session 1
Talking Cloud: Weekly AWS Round-up, AI in Focus and Exciting Announcements
มุมมอง 228 หลายเดือนก่อน
Talking Cloud: Weekly AWS Round-up, AI in Focus and Exciting Announcements
Exploring New IAM Access Analyzer Features: AWS reInvent Updates
มุมมอง 9711 หลายเดือนก่อน
Exploring New IAM Access Analyzer Features: AWS reInvent Updates
did you find a way to create index not manually
+1 I need really want to about it, too.🤔
+1 I also want to know how to create VectorIndex from CFT or some other way, not manually
Nothing yet, but let me look again. CFN changes so fast! :)
hi! thanks for the insightful video! any chance you have some of this in a Repo to check code?
Sorry, no. I am considering setting something up though
Thank you it was really helpful!
Glad it was helpful!
Straight to point thank you
Glad you liked it.
These are awesome, can you create something with aws cdk
WIP :)
This is great content. Can you also create similar videos for aws cdk to spin up infras.
Yes, definitely
Hello, the content is very good, you need to work on Amazon Q with AWS RDS, do you have any material? I can communicate but I'm having trouble searching
Hey, I'm glad you enjoyed the content. I don't have anything for RDS. Let me poke around a bit and see what I can dig up.
How do you create a launch template so that instances aren't assigned a public IP in the first place?
Normally, I would do this at the subnet level rather than in the EC2 configuration. Here's a CloudFormation template to get you started with a launch template - github.com/curiousorbit-com/ec2-launch-template/blob/main/launch-template.yml Here's a CloudFormation template for a VPC where the subnets are configured to not automatically assign public IP addresses - github.com/curiousorbit-com/cloudformation/tree/main/services/vpc
Great video. Do you have source code on any public repo ?
Not yet. I'll fix that this week.
@@brettg98 kind reminder on public repo/ code samples
@@TomerShaiman thanks 😀I've been busy with work, it's on my list.
thanks for this, if I am running opensearch privately... is there an equivalent to this AWS "Knowledgebase" there?? is it simply loading the vectors into a vector DB?
Hey! I'm glad you liked it. You can deploy OpenSearch into a VPC. To communicate with it you'd need to deploy an VPC Interface endpoint. I'm planning on doing this in a future stream, but you can have a look at the AWS documentation here for some guidance - docs.aws.amazon.com/bedrock/latest/userguide/knowledge-base-setup.html
Great video, thank you for sharing!
Glad you liked it!
Great videos! Looking forward to the next one. We decided to move into AWS from on prem hosting and this has been invaluable.
Hey! I'm glad you enjoyed the video. I'm working on the next one now.
Great video. I just waded into AWS for AI stuff from doing it on local hardware. This was very helpful!
Thanks for the feedback! I'm working on the next part now.
ImageRecipe referenced by !Ref returns object ARN, which means !GetAtt recipe0.Arn is not needed. DependsOn is not needed if you use !Ref function against recipe0 object, in this case CF will treat it as implicit dependency for pipeline0 object. Same goes for AWS::ImageBuilder::InfrastructureConfiguration. Great video but I don't see the last 15min
Appreciate the feedback. I tend to use DependsOn more than needed. It's an old habit. I need to record this video again. Something happened at the end :)
Now how we can SFTP in server to make changes to any files or folders ?
If you need to get OS level access, consider using Systems Manager. Have a look at 'Fleet Manger'. It will let you open both RDP or SSH (console) sessions without needed to be on the public Internet.
how to use a DEFAULT prefix list in cloudformation? I am not able to find example of syntax of how to use pl-31a34658 - com.amazonaws.global.cloudfront.origin-facing to allow access from cloud front
Once the Web Experience is created is it possible to send a prompt to it through the Boto3 API and get a response..
I would suspect this is possible, but I haven't tried it myself. Let me poke around a bit and see what I can come up with.
@@brettg98 is it possible? I'm trying here but I can't get nothing
Fantastic video! I can't wait for Part 3 next Tuesday!
@nobiaaaa, I'm so glad you enjoyed the video! We'll see you on Tuesday!
It seems now we can only create bedrock kb in the web console. And the bedrock kb looks limited. Maybe just use opensearch serverless vector index and implement a real custom kb.
This is so good a tutorial I subscribed right away!
@nobiaaa, I'm so glad you liked it. Thanks for the sub!
Hey Brett really cool video. I do have one doubt. If we create a chatbot with amazon bedrock that can aswer questions from a knowledge base, will it be able to answer general queries like hi, hello etc. If not is there a library that can be used to handle general queries and direct them to the foundation model instead of the knowledge base to get a reply. Thank you in advance ❤️✨
Hey @jefinprince, that's a great question. Yes. Here's something from the AWS documentation - "Knowledge bases for Amazon Bedrock help you take advantage of Retrieval Augmented Generation (RAG), a popular technique that involves drawing information from a data store to augment the responses generated by Large Language Models (LLMs)." We should finish up the walkthrough next week, so we can verify that for you.
Hey Brett, Your video saved me time! Thank you so much!
Glad it helped!
Hello Brett, Thanks for the concise explanation. If I remove the public IP of an instance whose subnet is attached to an IGW. Would it still be able to access the internet? What would be its public IP in this case?
Hey, I'm glad you liked the video. No, if you remove the public IP address the instance would no longer be able to access the internet. At that point, it would only have a private IP address (assigned via the VPC) which is not routable.
@brerrg98 is there have any way you can launch number of Ec2 instance by passing count ?
Hey, the AWS::EC2::LaunchTemplate doesn't have a specific property to specify the number of instances to launch. You could do this by associating your Launch Template with an Autoscaling group, creating an EC2 fleet, through the management console or by making an API call. I hope that helps.
Hey man, what IDE are you using? Thanks!
Hey, I'm currently using PyCharm
Excellent Explaining.
Awesome video! Can you publish your code? It would be really helpful. Thanks!!
Working on getting something into GitHub!
Been trying to figure this out and like how to design my network its harder to wrap my head around than i thought i just want to host a few websites and apps for now but not sure how to create it so i fan scale later and then like just kinda confused on where to place what
Hey! In general, I'd suggest you build a VPC that spans two AZs. Within your VPC, create public, private and maybe protected subnets. Make sure you spread the subnets across the AZs. So you'd end up with 2x public, 2x private, and 2x protected subnets. One in each of your AZs. Public subnets = AWS services (think ALBs) Private subnets = Your web servers Protected subnets = DB (RDS) Instances Good luck!
@@brettg98 appreciate the reply bro! I'm still alil confused but at the same time get it lol I just cant find any references for what I want specifically....I prolly want to setup a bastion too so I feel like I'll prolly have to just play around with it all
@@comosaycomosah playing around is the best way. Have a look at the systems manager. It'll let you connect without needing a bastion. If you give me a bit more info on what you're trying to do maybe we could do a stream on it.
so with the oracle free tier you get 2 micro amd and 1 24gb arm or can split it up....not sure how elegantly i can explain it or even what i want 100% but def gonna try lol i started trying to describe it but will try this first...... i know i CAN with out a bastion but if you wanted a completely private connection (vpn/ippsec whatev) between where you ssh into the instance from (with ability to change to diff ips if i wanna use diff device) , didnt want the instance completely exposed online to be pounded all day by bots but you still wanted to host a website from a portion of it (if possible i think you can with a dynamic gateway or router? may be wrong tho)how would you go about it?.........basically i just want only the "website area" of the arm instance exposed to the public internet ....im almost there but still missing essential network knowledge somewhere when it comes to intricate things like this lol sorry for the choppy explanation too
@@comosaycomosah thanks for the additional info. We'll do a stream on this and post it here. I won't use Oracle Linux, but it should give you a good idea on how you could build the network for your webservers
cant see very much screen is blurry
Great video
Thanks!
Great video. Can you please share the code
Thanks a million mate, i needed to know this!
Awesome video for the beginner's like me
Did anyone else run into issues connecting the ec2 instance with the Kubernetes cluster. I type in “kubectl get all” and I get an error that I should check my ports for access privileges.
Hey Brett, you forgot to upload session #7 🙂Great videos by the way.
Hey Hi..I'm just looking for a cloud formation template for cloud watch alarm for SQS failed message using metric old-age messages. Can you please help me on this
Thanks
Awesome video but still freezing aroud 55 min mark
Pdf bro
I want this project broo
Nice
Glad you enjoyed it.
This is Gold !!
What is this programming application name? Please share it sir
Hey Antony, we're using YAML and AWS CloudFormation to do this work.
@@brettg98 code is yaml, which application you using, also subnet link please
@@antony2684 I use PyCharm as my IDE. It's overkill for what I do, but I'm so used to it, that changing seems like it would be a chore :) Also, I've made a few VPC templates available here: github.com/brettg98/vpc
Hi, can you please help to provide the github link for the code for session 1 and session 2
You can find the template for this demo here: github.com/brettg98/ec2-launch-template
Great video, can you show how I can launch that using "LaunchTemplateSpecification" in Cloudformation. I can't seem to get the syntax right on just this option. Type: 'AWS::EC2::Instance' Properties: LaunchTemplateSpecification: LaunchTemplateId: lt-02e9c9c0f4eda1a52 The reason being is we're unable to launch anything that's not IMDSv2 enabled, so I need to be able to modify existing Cloudformation Stacks to force the use of IMDSv2 on the instances.
Hey Chris, I'm glad you enjoyed the video. I'll create something and share it here.
Chris, I've made a new video with the explanation I think you're after: th-cam.com/video/ugAY68G9Dtc/w-d-xo.html Check the comments for a link to the template
@@brettg98 amazing ill check it out.
Hi Brett, This look THE best option but I am not so nerd like )) also 3 parts and almost 6 hours. do you have this all setup in a zip file so l can put it on AWS and unleash my Wordpress , lol can be a dream. Or do you have shorter course for stupid ppl like me that will allow me to Auto Scalable WordPress On AWS Cloud
Hello, any possible way you can assist me on this? Trying to get help from a friend. I’m not a tech person but he’s facing this error: InvalidBlockDeviceMapping: Volume of size 30GB is smaller than the snapshot-expect size >= 100GB. Trying install Openshift/Redhat cloud cluster. Again, sorry about the wording since I’m not at all a tech person. Any advice is appreciated. Thank you
Based on the error message it sounds like you are creating a new EC2 instance and the EBS volume is to small. I'd suggest making the volume size on your newly launched EC2 instance 100GB and see if that fixes the issue. Good luck, and let me know how it goes.
Hello, any possible way you can assist me on this? Trying to get help from a friend. I’m not a tech person but he’s facing this error: InvalidBlockDeviceMapping: Volume of size 30GB is smaller than the snapshot-expect size >= 100GB. Trying install Openshift/Redhat cloud cluster. Again, sorry about the wording since I’m not at all a tech person. Any advice is appreciated. Thank you
Based on the error message it looks like you're trying to create a new EC2 instance with a drive of 30GB, but the snapshot (backup) you're using is 100GB in size. When you create the new EC2 instance make the drive 100GB in size. Good luck!
It's still freezes at 55:24 :(
Really? OK, so we'll redo this in Twitch and get a new recording :)
@@brettg98 Hi do you have any news about the missing part of the video? Thx in advance
@Brett Gillett The Video was a good learn, do you have the missing part of the video, the one that freezes, also it would be great if you could share the script please
Pretty nice explanation 👌 thank you 😊
glad you liked it!