amazing work! thanks for the valuable info. content creators like you improve our developer lives so much more than the practically non-existent Microsoft documentation... hopefully it will get better in time
Great video 👍👍👍 Just three questions: 1] There is already a new beta out (Version="2.1.0-beta.1" ) that breaks all code (20 errors) 2] How will this work if we need to upload multiple pdf's and do a search that works through all the uploaded files? 3] How can we do this using just OpenAI and not use AzureOpenAI? Thx and keep up the good work 🙂
I have put an "issue" and pull request on the repo showing how to do this with this version. It creates the vector store and then attaches it to the Assistant.
Ah mate youre a life saver this is exactly what I need. Couple of questions... So for instance I want to use this in a mvc or blazor web app and I have multiple users accessing the assistant at the same time, whould it treat each connection as a separate session? Would it be better queued and managed by a triggered azure function rather?
thanks for your work on this. In your last video you used a key and said it can be either azurekey or openAI key. With this code it looks like I need a URI from an azure registration, is this correct?
@@edandersen I have got an azure subscription and my code now works to a point. Like others I have a call response to a gui. I have a class containing the assistant, thread and file value Ids as well as filename, question and answer, so when asking a new question I am getting as far as await client.CreateMessageAsync(... which looks fine, but the resulting answer to the question is always null. Sounds like you have cracked this stage.
@@prmagic5941 They've updated the libraries since making the video sadly and I'm no longer focussed on this. Maybe ask an AI MVP on twitter if you can find one?
Thanks Ed. I downloaded the code and run locally but no output is being displayed. I can see the file was uploaded in the Azure OpenAI Service (under Data Files). It returns an empty line..and the message 'Your response (leave empty to quit)". And thats all.
@@edandersen I've read that you're on to something related to it, I'm looking forward to that :) I'm having a hard time calling the functions xD been trying to :3
@@edandersen the small company I work at is simply using azure as a secure way to use gpt-4. We don't have time to build our own interface. I wish I could use olllama...
@@edandersen the .pdf file is saved in the OpenAI Data Files section, but I’m not receiving any response or confirmation regarding this file, unlike what you experienced.
For an interactive debugging session you can point to the file you want in your launch.json file. For the args parameter you could include an entry like: ["${workspaceFolder}/files/"] where I created a project folder under the root named "files"
I’ve only just found your channel, and I’m loving the content. This one was really useful and has kicked off some ideas. Thanks!
I like these videos explaining the updates technologies
amazing work! thanks for the valuable info. content creators like you improve our developer lives so much more than the practically non-existent Microsoft documentation... hopefully it will get better in time
This is freaking awesome. Thank you for making this content.
Great Video! Please keep these coming!
It's hard to keep up with these libraries changing faster than I can develop a product. This helped me migrate to V2, thanks!
glad to have helped!
Great video 👍👍👍
Just three questions:
1] There is already a new beta out (Version="2.1.0-beta.1" ) that breaks all code (20 errors)
2] How will this work if we need to upload multiple pdf's and do a search that works through all the uploaded files?
3] How can we do this using just OpenAI and not use AzureOpenAI?
Thx and keep up the good work 🙂
1) Oh well, nothing I can do
2) You can attach multiple files the assistant session
3) The SDK should support both
I have put an "issue" and pull request on the repo showing how to do this with this version. It creates the vector store and then attaches it to the Assistant.
Nice work! Thanks for sharing.
No probs grab the code off the repo :-)
Thanks for creating a video on Assistants. Is it possible to implement Function Calling in C# for ADO integration?
Yes it is. I've done it and it works very well.
thanks for the video, any chance for an example with function calling ? being struggling with for quite a while, the docs seems quite confusing
I have got this working so I might just do that
@@edandersen looking forward to this!
Ah mate youre a life saver this is exactly what I need. Couple of questions... So for instance I want to use this in a mvc or blazor web app and I have multiple users accessing the assistant at the same time, whould it treat each connection as a separate session? Would it be better queued and managed by a triggered azure function rather?
I’d just set up an assistant thread per user
thanks for your work on this. In your last video you used a key and said it can be either azurekey or openAI key. With this code it looks like I need a URI from an azure registration, is this correct?
best check the docs - they've changed it all again.
@@edandersen I have got an azure subscription and my code now works to a point. Like others I have a call response to a gui. I have a class containing the assistant, thread and file value Ids as well as filename, question and answer, so when asking a new question I am getting as far as await client.CreateMessageAsync(... which looks fine, but the resulting answer to the question is always null. Sounds like you have cracked this stage.
@@prmagic5941 They've updated the libraries since making the video sadly and I'm no longer focussed on this. Maybe ask an AI MVP on twitter if you can find one?
Thanks Ed. I downloaded the code and run locally but no output is being displayed. I can see the file was uploaded in the Azure OpenAI Service (under Data Files). It returns an empty line..and the message 'Your response (leave empty to quit)". And thats all.
Things have changed since I made this video, if you can figure out what has changed let us know :-)
Hi! Do you have a video for Azure OpenAi Assistants function calling in c#?
Hi, not yet no. I have done it though and its pretty cool.
@@edandersen I've read that you're on to something related to it, I'm looking forward to that :) I'm having a hard time calling the functions xD been trying to :3
Is there a nicer front end available than just the playground?
I think you need to build it
@@edandersen the small company I work at is simply using azure as a secure way to use gpt-4. We don't have time to build our own interface. I wish I could use olllama...
Instead of uploading/deleting file everything, for big files It make sense to upload once and don't delete, possible?
yeah absolutely. I just do it to clean up.
Biggest problem I have not adopted the azure openai is because of the examples in python.
why this is not warking for me?
make sure assistants are available in your azure open ai region. also check the source code linked in the description
@@edandersen the .pdf file is saved in the OpenAI Data Files section, but I’m not receiving any response or confirmation regarding this file, unlike what you experienced.
@@HakobAmirjanyan did you solve the issue? Faced with the similar one
I've cloned your repo, just crashes with null ref here:
var filename = args[0];
you need to provide a filename as the first argument
For an interactive debugging session you can point to the file you want in your launch.json file. For the args parameter you could include an entry like:
["${workspaceFolder}/files/"]
where I created a project folder under the root named "files"