marvin VS ctop

Hey HN! We just released Marvin 2.0.

Marvin is an AI toolkit for developers who want to use LLMs with traditional software. We still see significant challenges integrating LLMs because of how difficult it is to get them to reliably accept and return structured data. Marvin consists of independent, functional tools that address this problem in a variety of ways.

Marvin has always been focused on using LLMs to work with native Python datatypes and Pydantic models. In 2.0 we've expanded this significantly with dedicated APIs for the most common use cases we've seen over the last year: classification, entity extraction, transforming data to types, and generating synthetic data. Marvin 2.0 is also fully multi-modal and supports images as inputs for classification, extraction, and transformation tasks (as well as simple image and speech generation). We've also introduces a Pythonic interface to OpenAI's assistants API, which now powers all of Marvin's interactive components.

We've tried to make an LLM framework that "sparks joy" and captures that same feeling you had the first time you saw an LLM in action. Try it out and let us know what you think!

(Repo: https://github.com/PrefectHQ/marvin)

Seems a lot like https://github.com/PrefectHQ/marvin?

The prompting you do seems an awfully like:

https://www.askmarvin.ai/prompting/prompt_function/

You can try the decorator ai_fn in marvin https://github.com/PrefectHQ/marvin

Marvin: a batteries-included library for building AI-powered software. Marvin's job is to integrate AI directly into your codebase by making it look and feel like any other function (https://github.com/PrefectHQ/marvin)

Sure! I was inspired by this Python library: https://github.com/PrefectHQ/marvin

I see Langchain has support for Azure chat models, and Marvin is built on Langchain so it may not be so difficult! Tracking issue here: https://github.com/PrefectHQ/marvin/issues/189

We have a related issue open (https://github.com/PrefectHQ/marvin/issues/64) but haven't designed anything yet.

This does remind me of ctop as well: https://github.com/bcicen/ctop

It also let's you look at containers, resource usage graphs, their logs and even do some actions through a TUI.

ssh, nnn, micro and ctop is all I need on my dockerhosts

My problem with ctop is, that it seems to show wrong memory usage data: https://github.com/bcicen/ctop/issues/314

When talk about interface and cli... I am a huge fan of ctop

You can use CTOP. It's like a lite portainer on CLI. You can check logs, stats, restart containers.

In the terminal, there are also a few useful projects:

  - for Docker, there is ctop: https://github.com/bcicen/ctop

> I tried portainer, awful UX experience and all good features are inside paid version.

This is interesting to me, because it doesn't quite match my experience - I've been using Portainer for around 3 years at this point and it's been pretty decent.

The worst issues that I've gotten is networking issues in some hybrid configurations with Docker Swarm (e.g. Portainer cannot reach the manager node of the cluster for a bit), or troubles configuring Traefik ingresses when managing Kubernetes (though I think the recent patch notes talked about improving the ingress section, so maybe the experience will get better with non-Nginx ingresses).

Other than that, it's been great for onboarding new people, illustrating the cluster state at a glance, easily operating with stacks and scaling/restarting services as needed, including pulling new images, viewing the logs or even connecting to containers through a web UI if need be. The webhook functionality in particular is really nice - you can just do a curl request against a given URL and that will pull the new container versions for the given image and do a redeploy, which works nicely with a variety of CI solutions.

When I last tried, initializing Nomad clusters with networking encryption was a bit less of a smooth experience (needing to essentially manage your own PKI) and the web UI felt more like a dashboard, instead of something that you could click around in, if you're a proponent of that workflow.

Rancher is probably better than both of those options, though there's a certain overhead in regards to running both that software and a full Kubernetes cluster. If Kubernetes feels like a good fit for a particular project and resources aren't an issue, definitely check it out! You can, of course, also have some success with lightweight clusters, like K3s: https://k3s.io/

I'll definitely agree that Lazydocker is a nice tool, but I wouldn't call it superior, just different (TUI vs GUI), their demo video is nice though: https://youtu.be/NICqQPxwJWw

It actually reminds me of ctop, which you might also want to check out, though it's not something that you'd manage clusters in, merely the individual containers on a node (which won't always be enough, same as Docker Compose isn't): https://github.com/bcicen/ctop

Regardless, for Kubernetes, I'm inclined to say that you'd enjoy k9s a bunch then, it has a similar TUI approach: https://k9scli.io/

However, something like ctop may be easier to use.