5 minutes ago, Reptar said:

I noticed this error:

Failed to update API costs: KeyError: 'gpt-3.5-turbo-0125'

 

Have you checked more into it? Is the key used correct? Maybe the model should be something else?

 

Unfortunately I stopped messing around with autogpt and I don't specifically recall running into issues on the last docker image of autogpt_gotty

 

I'm not sure what I would need to do to resolve that error. I have entered the correct OpenAI API key and have billing enabled on my account.
 

Regarding "NotImplementedError: The Redis memory backend has been rendered incompatible by work on the memory system, and has been removed temporarily.", do you think this could be a simple fix with how I have redis configured? I'm not sure what I'm doing wrong but am hopeful it's something straightforward I don't know about yet

8 minutes ago, icarus69 said:

I'm not sure what I would need to do to resolve that error. I have entered the correct OpenAI API key and have billing enabled on my account.
 

Regarding "NotImplementedError: The Redis memory backend has been rendered incompatible by work on the memory system, and has been removed temporarily.", do you think this could be a simple fix with how I have redis configured? I'm not sure what I'm doing wrong but am hopeful it's something straightforward I don't know about yet

 

Maybe make sure the correct model is listed then? Maybe openai deprecated the one you're trying to use?

 

On the redis thing, the latest docker image had a version of autogpt that had that disabled by the actual autogpt project. There's nothing you did wrong there, they just temporarily disabled it on that release.

9 minutes ago, Reptar said:

 

Maybe make sure the correct model is listed then? Maybe openai deprecated the one you're trying to use?

 

It looks like that is the latest version of 3.5 turbo, so openai clearly haven't deprecated it.

 

Quote

There's nothing you did wrong there, they just temporarily disabled it on that release.

 

How do I re-enable it and/or which image can I change/revert/update to that's out there, or resolve the issue in some other way?

2 minutes ago, icarus69 said:

It looks like that is the latest version of 3.5 turbo, so openai clearly haven't deprecated it.

 

How do I re-enable it and/or which image can I change/revert/update to that's out there, or resolve the issue in some other way?

OK so I found this issue that has that same error: https://github.com/Significant-Gravitas/AutoGPT/issues/4887

 

My bet is that something changed in the API from openAI and the outdated version that is on the autogpt_gotty image is incompatible. I'm making this assumption because the guy in the issue mentioned that the error went away once he updated autogpt.

 

I can try to update the docker image but I'm not sure when that might be.
You can also build a new image using the docker file I provide here: https://github.com/ther3zz/autogpt_gotty/blob/main/Dockerfile

That should fix things (assuming the autogpt project didnt change directories too much etc)

2 minutes ago, Reptar said:

OK so I found this issue that has that same error: https://github.com/Significant-Gravitas/AutoGPT/issues/4887

 

My bet is that something changed in the API from openAI and the outdated version that is on the autogpt_gotty image is incompatible. I'm making this assumption because the guy in the issue mentioned that the error went away once he updated autogpt.

 

I can try to update the docker image but I'm not sure when that might be.
You can also build a new image using the docker file I provide here: https://github.com/ther3zz/autogpt_gotty/blob/main/Dockerfile

That should fix things (assuming the autogpt project didnt change directories too much etc)

 

 

Gotcha. Do you think that would also resolve the redis issue? I also wouldn't be surprised if there's some config I've got slightly wrong with that (e.g. ports, mapping, directory) but maybe not.

 

I don't have experience with Docker until recently and definitely don't know how to build an image, but if it's not too difficult I could try.

3 minutes ago, icarus69 said:

 

Gotcha. Do you think that would also resolve the redis issue? I also wouldn't be surprised if there's some config I've got slightly wrong with that (e.g. ports, mapping, directory) but maybe not.

 

I don't have experience with Docker until recently and definitely don't know how to build an image, but if it's not too difficult I could try.

Yeah, I would assume the autogpt project would have re-enabled redis by now so I'm betting it should work. I quickly checked your mappings and it all looks correct, I didnt really see anything wrong there.

 

I didnt have any experience either until I started messing with nextcloud and then jumped into the autogpt_gotty thing! It's definitely come in handy since then though, I definitely recommend you try it out!

It took me a couple of days to understand it well enough to push out that image but in the end it was worth it for the experience!

1 minute ago, Reptar said:

Yeah, I would assume the autogpt project would have re-enabled redis by now so I'm betting it should work. I quickly checked your mappings and it all looks correct, I didnt really see anything wrong there.

 

I didnt have any experience either until I started messing with nextcloud and then jumped into the autogpt_gotty thing! It's definitely come in handy since then though, I definitely recommend you try it out!

It took me a couple of days to understand it well enough to push out that image but in the end it was worth it for the experience!

 

Appreciate the encouragement!

 

Any good tutorials online that can show me how to create an image? Not sure what the step by step is here from the point of taking your git project.

5 minutes ago, icarus69 said:

Appreciate the encouragement!

 

Any good tutorials online that can show me how to create an image? Not sure what the step by step is here from the point of taking your git project.

Take a look at this post: 

I shared the sources I used to build the image initially

 

Also, dont forget to use chatgpt as a resource, it helped me out too!

Of course feel free to reach out here as well and I can try to help where I can! 

 

15 minutes ago, Reptar said:

Take a look at this post: 

I shared the sources I used to build the image initially

 

Also, dont forget to use chatgpt as a resource, it helped me out too!

Of course feel free to reach out here as well and I can try to help where I can! 

 

 

Thanks, I think I understand vaguely what needs to be done.

 

In much simpler terms though, I really don't know what I'm meant to do to get started. Create a fork from your project? From the original? And why are there so many fewer files in your project compared the the main original autogpt? Do I need to have all the files from the original, or just yours? And then should I be creating and editing this image from my local device in terminal, or in Unraid?

Did you ever get this working? I'm interested in getting private-gpt working in Unraid, surprised there isn't a Community App for this yet.

52 minutes ago, rexkwondo said:

Did you ever get this working? I'm interested in getting private-gpt working in Unraid, surprised there isn't a Community App for this yet.

I switched to using open webui and ollama. Haven't turned back since!

  

13 hours ago, Reptar said:

I switched to using open webui and ollama. Haven't turned back since!

Thanks for the suggestion, I hadn't even heard of that. Did you use a community app and if so, which one do you recommend? I see there are a few of them here:

 

18 hours ago, rexkwondo said:

  

Thanks for the suggestion, I hadn't even heard of that. Did you use a community app and if so, which one do you recommend? I see there are a few of them here:

 

I used the official Ollama image listed in that image.

 

I also used an old app listed as ollama webui which has since been deprecated. I ended up grabbing that template and changed the repo to ghcr.io/open-webui/open-webui:latest

 

Check out their envs here: https://docs.openwebui.com/getting-started/env-configuration/

 

 

Let me know if you have any other questions, I've been pretty involved with open-webui!