backend-dev-33

I'm backend developer and have to build a frontend for my project. Can write simple JS, but would avoid Big Javascript Frameworks ))

This should be an almost static site:

- some pages will contain a kind of custom search component: an input field with 10-12 checkboxes/dropdowns containing HTML+JS+CSS. already have a working prototype.

- other pages like About/Contact/FAQ/Help - completely static, pure Bootstrap HTML/CSS (and minimal JS)

Question1: suggest a TEMPLATE ENGINE. Something similar to Jekyll would be great (used Jekyll in the past - the template system is OK, but not the Ruby parts of it). Something that has good integration with Bootstrap and Liquid templates

Question2: suggest a JAVASCRIPT BUNDLER. Should have good integration with template engine and Bootstrap. Probably not Webpack: I'm afraid of those huge config files. Tried Parcel a bit: it is not bug-free, the experience was not smooth. Don't know about Vite.

Question3: what is known about usage of BOOTSTRAP (+TEMPLATE ENGINE) WITH AI CODE EDITORS ? (Cursor, Windsurf or something else) I've heard stories of people generating big chunks of applications with these things. I think it should work well with Bootstrap HTML, but I don't know how it would work with the template engine.

I want to do zero-shot text classification with this model [1] or with something similar (Size of the model: 711 MB "model.safetensors" file, 1.42 GB "model.onnx" file ) It works on my dev machine with 4GB GPU. Probably will work on 2GB GPU too.

Is there some hosting provider for this?

My app is doing batch processing, so I will need access to this model few times per day. Something like this: start processing do some text classification stop processing Imagine I will do this procedure... 3 times per day. I don't need this model the rest of the time. Probably can start/stop some machine per API to save costs...

[1] https://huggingface.co/MoritzLaurer/roberta-large-zeroshot-v2.0-c

I want to do zero-shot text classification either with the model [1] (711 MB) or with something similar. Want to achieve high throughput in classification requests per second. Classification will run on low-end hardware: some Hetzner [2] machine without GPU (Hetzner is great, reliable and cheap, they just do not have GPU machines), something like this:

* CCX13: Dedicated vCPU, 2 VCPU, 8 GB RAM

* CX32: Shared vCPU, 4 VCPU, 8 GB RAM

Now there are multiple options for deploying and serving LLMs:

* lmdeploy

* text-generation-inference

* TensorRT-LLM

* vllm

There are more and more new frameworks for this. I am a bit lost. Would you suggest the best option for deploying the above-listed model (No-GPU hardware)?

[1] https://huggingface.co/MoritzLaurer/roberta-large-zeroshot-v2.0-c

[2] https://www.hetzner.com/cloud/

I am looking for software to classify documents into 10-20 categories. The documents are about half-screen to screen long.

There are some labeled data (about 50-80 labeled documents per category. not 500 per category), so a few-shot learning might be an option.

Algorithms used: it might be something like KNearestNeighbor or some ML/Neural networks (transformers? LLM?). Should just do the proper classification.

Some restrictions: It should be a "ready to use" pipeline with documentation about training the model, parameter optimization etc. If possible - there should be some way to use this framework/library without Python (I'm not a Python developer) For example, the [1] and [2] allow to use command-line interface for everything - it seems using Python is optional for these frameworks. The SetFit framework (see [3] and [4]) looks quite promising (good results with 8 labeled samples per class!). But requires doing everything in Python.

[1] https://fasttext.cc/docs/en/supervised-tutorial.html

[2] https://neuml.github.io/txtai/pipeline/text/labels/

[3] https://github.com/huggingface/setfit

[4] https://www.philschmid.de/getting-started-setfit

Part1: https://news.ycombinator.com/item?id=20256681 (closed to new comments)

What (legal?) trick allows search engines to crawl(well, we know that "crawl" is synonim of "scrape") and index content protected by terms of use? Is it "fair use" or something else?

One example: Craigs List!

In their terms of service:

> USE. Unless licensed by us in a written agreement, you agree not to use or provide software (except general purpose web browsers and email clients) or services that interact or interoperate with CL, e.g. for downloading, uploading, creating/accessing/using an account, posting, flagging, emailing, searching, or mobile use. You agree not to copy/collect CL content via robots, spiders, scripts, scrapers, crawlers, or any automated or manual equivalent (e.g., by hand).

On the other hand: https://www.google.com/search?q=site%3Asfbay.craigslist.org+couch&oq=site%3Asfbay.craigslist.org+couch

Google is able to index CL and you can query the google index specifying "use only this CL city" and you can see the ads, and we know Google making money with it (advertising for example).

I can not imagine google obtaining "written agreement" from CL ))

For a static Github Pages website, I would like to have a commenting system. The idea is to use Firebase, because: 1. It gives social login: Google, Twitter, Github, FB 2. I want to use Firebase for some other functionality not related to comments. For a user it should be very convenient: once authenticated he/she can comment and can use other site features (these "features" rely on Firebase)

So much theory. I started to search for Firebase commenting system and... I can find only the projects with 0 or 1 release, projects with 1 contributor, projects abandoned 5 years ago, example projects... you got the idea.

A commenting system should have a few features: something like an admin interface; something to fight/prevent spam. I can not find such a thing with Firebase as a backend!

There are quite advanced self-hosted systems, Coralproject [1] and Commento [2] are good examples. They do not use Firebase.

I do not need all the features of Coralproject. But a commenting system with Firebase backend, antispam and with something like an admin interface would be great. Do you know such a thing?

[1] https://github.com/coralproject/talk [2] https://gitlab.com/commento/commento/

For years I used Hetzer VQ12 (unmanaged virtual machine with root, Single Core, 1024 MB RAM, 40 GB HDD) for experiments and to run some Java processes.

Nothing special, nothing fancy, just a relatively cheap (12,90 € per month) and relatively reliable (one or two times per year notifications about scheduled downtime for about 15-30 min) hardware.

At the moment there is a cronjob on this machine. Every hour it triggers a Java process which runs for about 5-10 minutes.

Now Hetzner going to discontinue these servers (they notified me in advance) so I want to move either to another Hetzner product (probably their "cloud" CX11 or CX21) or to another hosting provider.

Please share your experience. I heard some positive feedback about digitalocean (droplets?)

UPDATE: I do not want to have dependencies on cloud infrastructure. Just plain old Linux command line, nothing more

I'm building a website. It is bigger than just a side project. Found an image on one of the stock image sites, bought it for $20. I use this image as site a logo. I contacted the artist. Asked him to make few illustrations for my website in the same style. He did. Everything fine so far.

No I'm thinking about uniqueness of this logo image. The picture was sold for a long time on this (and few others) stock image sites. It is still available there.

Should I get (buy) some exclusive rights ? If I do this — what happens to the people who bought this image before? Should I just add the unregistered trademark™ symbol somewhere near this image and site name? Should I ask the artist to modify the image to make it unique? (something like adding one or two letters from the site name, "pied piper" style, you know))