ChatPad on Cloudron - ChatGPT Artificial Intelligence Interface

[marcusquinn]

@LoudLemur Got another one for you:

• https://forum.cloudron.io/topic/8872/serge-llama-made-easy-self-hosted-ai-chat

I think we'll see a lot more. What an amazing time to be alive!

[LoudLemur]

@marcusquinn said in ChatPad on Cloudron - ChatGPT Artificial Intelligence Interface:

@LoudLemur Got another one for you:

• https://forum.cloudron.io/topic/8872/serge-llama-made-easy-self-hosted-ai-chat

I think we'll see a lot more. What an amazing time to be alive!

That one is fantastic! All credit to those researchers who on a tiny budget gave the world such incredible Free Software. Please upvote that request, if anybody is reading this.

[timconsidine]

@LoudLemur I noticed there is a Desktop version of the app.
Saves building a Cloudron app !
But will look at that.

[robi]

@timconsidine does it allow to connect to locally running GPTs?

[timconsidine]

@robi hmmm, don't see that functionality

[robi]

@timconsidine I added that idea over at Reworkd.ai's https://github.com/reworkd/AgentGPT/issues/322

[Kubernetes]

I am working on a package for ChatPad... (it's already working on my Cloudron, just need to push it to github)

[Kubernetes]

Done, the package is ready and running fine on my Cloudron Instance.
The package can be found in Gitlab

As this is my second package, and I didn't get yet Feedback for my first package, I am asking for feedback, if I did that packaging the right way

[timconsidine]

@Kubernetes said in ChatPad on Cloudron - ChatGPT Artificial Intelligence Interface:

I am asking for feedback

overall format looks good, but for an 'adopted' Cloudron app, Dockerfile FROM should be the Cloudron base image
FROM cloudron/base:4.0.0@sha256:31b195ed0662bdb06a6e8a5ddbedb6f191ce92e8bee04c03fb02dd4e9d0286df

and then do something to pull in the code from the source repo (but there are other ways)

ADD https://github.com/CorentinTh/it-tools/archive/refs/heads/main.tar.gz /app/code
RUN tar -xzf main.tar.gz && rm main.tar.gz && mv it-tools-main/* . && rm -r it-tools-main

But maybe others with more experience and authority can advise otherwise.

For a custom app which is only to be used by author or small group, if it works, it works.
For a custom app which might be adopted with refinements into the App Store, I think it is better to take the above approach.

[LoudLemur]

@Kubernetes said in ChatPad on Cloudron - ChatGPT Artificial Intelligence Interface:

Done, the package is ready and running fine on my Cloudron Instance.
The package can be found in Github

As this is my second package, and I didn't get yet Feedback for my first package, I am asking for feedback, if I did that packaging the right way

Thank you for doing that.

[Kubernetes]

@timconsidine Thanks for the feedback. I didn't notice that I should use the cloudron base image. I will try to it in the work I already have done and update then.

[Kubernetes]

Okay, I have done it. The package is ready from my point of view and could be published in Cloudron App Store.

Thanks to all for the feedback and support.

[timconsidine]

Will take a look later. Thinking/hoping it will be better than my self-package of the other one (chatbot-ui).

I don't suppose you feel up to the challenge of packaging Flowise ?
https://forum.cloudron.io/topic/9130/flowise-ui-for-langchain?_=1685224667920

Working with regular ChatGPT, even V4, the knowledge cut-off date is sometimes problematic or time-wasting.
Being able to include more current resources (like developer updated docs) is likely to be super helpful.
or so I hope

[Kubernetes]

@timconsidine Interesting tool, I tried, but first attempts did fail because of the read-only filesystem in Cloudron...

[timconsidine]

@Kubernetes I am trying also but running into errors
Will try again later

[timconsidine]

@Kubernetes said in ChatPad on Cloudron - ChatGPT Artificial Intelligence Interface:

because of the read-only filesystem in Cloudron

Shhhh, don't tell @girish or @nebulon but sometimes I try to "bypass" this issue to get a proof of concept by installing to /app/data
Just to see it if builds and works.
Then think about separation of code and data

[timconsidine]

@Kubernetes said in ChatPad on Cloudron - ChatGPT Artificial Intelligence Interface:

Okay, I have done it. The package is ready from my point of view and could be published in Cloudron App Store.

Well done
I have a built the package, it installed smoothly and runs nicely.
Thank you for your work on this. I may stop using the desktop version.

The Docker image is large : 5Gb. Surprised me (but heck, what do I know).
I wonder if there is a way to reduce the size of the image during the build process.
But I am no expert in Docker, so not sure.
I didn't see anything obvious in the Docker file.
Maybe a multi-stage build ? But I don't much about that.

[Kubernetes]

I moved the repo to the Cloudron Gitlab

Regarding size, 5Gb is really big. But maybe someone with experience in shrinking the Docker images can give some advise.

[Kubernetes]

I have updated the Dockerfile to remove unrequired software and use of multi stage for build and production.

[timconsidine]

@Kubernetes oooo ! Will take a look