Closed samleegithub closed 1 year ago
This sounds like an interesting idea. I'll go ahead and add this to our backlog of samples that we should create.
@samleegithub - this is interesting, thanks for bringing this up. Can you create a PR that creates a Plugin to do this.
@evchaki - I was thinking the demo might be a sample dotnet notebook that uses stepwise planner, bing search skill, web-scraper plugin and potentially azure cognitive search for memory.
Were you thinking that all of this would be wrapped up inside a researcher plugin?
I started an implementation of this last week.
While there isn't much on the repo yet, the researcher code is pretty much what @samleegithub mentions above:
The webscraper plugin is currently giving me some headaches deploying to Azure, it's working locally right now: https://github.com/craigomatic/webscraper-aiplugin/
There might be 1 or 2 more plugins that we create to pull data from less general places than a Bing search, although I'm not sure what those would be just yet (maybe searching relevant arvix papers?)
@samleegithub let me know if you'd like to join forces,
@craigomatic Awesome! I'm not an expert C# coder but I can build and run the code to test.
I tried to install the webscraper plugin on my local machine, but I'm getting this error:
dotnet run
Unhandled exception. System.InvalidOperationException: The gRPC channel URI 'http://:7071' could not be parsed.
at Microsoft.Azure.Functions.Worker.Grpc.GrpcWorkerClientFactory.GrpcWorkerClient.CreateClient() in D:\a\_work\1\s\src\DotNetWorker.Grpc\GrpcWorkerClientFactory.cs:line 110
at Microsoft.Azure.Functions.Worker.Grpc.GrpcWorkerClientFactory.GrpcWorkerClient..ctor(GrpcHostChannel outputChannel, GrpcWorkerStartupOptions startupOptions, IMessageProcessor processor) in D:\a\_work\1\s\src\DotNetWorker.Grpc\GrpcWorkerClientFactory.cs:line 52
at Microsoft.Azure.Functions.Worker.Grpc.GrpcWorkerClientFactory.CreateClient(IMessageProcessor messageProcessor) in D:\a\_work\1\s\src\DotNetWorker.Grpc\GrpcWorkerClientFactory.cs:line 33
at Microsoft.Azure.Functions.Worker.GrpcWorker.StartAsync(CancellationToken token) in D:\a\_work\1\s\src\DotNetWorker.Grpc\GrpcWorker.cs:line 57
at Microsoft.Azure.Functions.Worker.WorkerHostedService.StartAsync(CancellationToken cancellationToken) in D:\a\_work\1\s\src\DotNetWorker.Core\WorkerHostedService.cs:line 25
at Microsoft.Extensions.Hosting.Internal.Host.StartAsync(CancellationToken cancellationToken)
at Microsoft.Extensions.Hosting.HostingAbstractionsHostExtensions.RunAsync(IHost host, CancellationToken token)
at Microsoft.Extensions.Hosting.HostingAbstractionsHostExtensions.RunAsync(IHost host, CancellationToken token)
at Microsoft.Extensions.Hosting.HostingAbstractionsHostExtensions.Run(IHost host)
at Program.<Main>$(String[] args) in C:\Users\samulee\Documents\Microsoft Projects\ITF\webscraper-aiplugin\Program.cs:line 52
Browsing through the code, it looks like I'm missing a local.settings.json file.
Good catch, readme updated and empty local.settings.json was added to the repo.
@craigomatic I updated to the latest code and filled in the config values.
But I'm still getting the same error saying https://:7071 could not be parsed. I can't figure out how to fix this.
You might be missing the Azure Function tools: https://learn.microsoft.com/en-us/azure/azure-functions/functions-develop-local#local-development-environments
I figured out what I was doing wrong. I was using "dotnet build" and "dotnet run". I should have been using "func run" instead.
It seems to be working now. I'm testing with Chat Copilot now.
Thanks!
Closing this issue since there hasn't been activity in awhile. @samleegithub, feel free to reopen this ticket if you end up submitting a PR for a notebook example.
Inspired by this video: https://www.youtube.com/watch?v=ogQUlS7CkYA
https://github.com/JayZeeDesign/researcher-gpt
They create an AI Agent that is able to do research automatically using LangChain.
Would it be possible to create a similar demo using Semantic Kernel instead of LangChain?
I imagine it would be a combination of Stepwise Planner, Azure Cognitive Search (vector search) and a web-scraper plugin.