A small locally run model would work and it would use less electricity than what's needed for spending time researching different lists from different websites.

Sure, maybe LLMs would find this challenging, but the same is true for humans. There is never going to be a second human who is better at this than Claude, like 70%? Jeeeze..

What is your small lm threshold?

There are Gemma 4 models that run on phones, is that what you’re looking for?

To do this well you will need tool calls because just reading the git diff won’t be enough context to do a good job.

This is probably something you can fine tune, it is getting exponentially cheaper to do this, but without knowing why you don’t want to self host an existing LLM to solve this problem it’s hard to give good advice.

You can look at smollm3 or nanogpt and nanollama as (dated) examples of how to build such a small model yourself, but they’re still LLMs that run on LLM tools. Smollm has guides to fine tune for your task.

You could run them with PyTorch directly or something too, if you’re opposed to the full llama.cpp (vllm is even heavier, with very long startup times).

I’ve met a lot of people who previously worked at FAANG, or whatever the acronym is these days, they were not all smart. Actually, most of them couldn’t use a French Press (I exaggerate). Their intelligence comes in creating algorithms that accomplish more capitalism. Outside of the math and the logic there isn’t a lot there.

I mean, that one line summary is definitely doable for an LLM (not so sure for a small model, but maybe with the right harness).

The only reason I wouldn't consider it useful there, or perhaps even harmful, is that I can really recommend knowing the gist of that summary long before you think to commit. You should have an idea of what atomic change you want to make before you start coding.
Obviously, you'll likely be wrong and need to make other changes beforehand, but then you can git stash, make the other change and continue on your original objective.

I've had juniors before that really struggled with not getting overwhelmed trying to change everything all at once, and it really clicked for them when I told them to think of their commit message before they start, so yeah.

Having said that, I have a colleague that's not too hot with English and probably also has dyslexia, and he just writes the rough message, then has a language model clean up the formulation + typoes. I do think that's a good use for a language model.

Well, I agree that they shouldn’t use this name. That’s not even a mermaid right? I have no problem with them using a Hawaiian mermaid name for a markdown mermaid app.

I don’t agree with the name usage. Mustang is a Horse, isn’t should t be used outside of the equestrian scene? What about the white gods, I bet you’re fine using those.

Pick your favorite:

Nike, Hermes, Apollo, Zeus, Athena, Ares, Artemis, Poseidon, Hades, Hera, Atlas, Titan, Triton, Eros, Helios, Hyperion, Pegasus, Pandora, Dionysus, Aphrodite, Hercules, Mercury, Achilles, Morpheus, Nemesis, Iris, Selene, Demeter, Prometheus, Kronos, Gaia, Pan, Echo, Nyx, Orion, Phoenix, Ajax, Argus, Olympus, Perseus, Medusa

There’s an Orion web browser…. I’ll bet you’ve got some words for them.

“Meaningful” is doing the heavy lifting.

To make a good commit message it must understand your codebase, and programming in general.

lolwut

A commit message is not supposed to be a summary of the code. It is supposed to gives information that you cannot deduct from the diff. Like:

• Is there an attached ticket ?
• Does it produce side effects ?
• Is it a quick and dirty fix just before the release ?
• What are the consequence of the modification ?
• Why you choose this approach instead of another ?
• Why are you not able to add regression test ? etc ...

The first guy at Cursor that create the feature to generate commit message from diff is just a fucking moron that don't understand the role of a commit message.

Well, if a meaningful commit message can be created only with a model as big as 1B parameters, I'm better off doing them meself.

You might not want to admit it, but you’re asking for a local LLM.

“Small” is just a euphemism for local.

This man should get into political extremism. He's clearly unwell and I find those types tend to be the best targets to manipulate into performing extremist acts.

Just saying someone should get around to honey-potting him. He's clearly asking for it.

surely you're joking.

The premise was that this language is significantly different from other languages in terms of being understood and replicated by an LLM.

The result showed that an LLM does exactly as well with it as you would expect.

Commit messages are for sure short and easy to come up with, already. The problem's typing some common stuff every single time.

Thanks for letting us know. Yeah sure. It's not worth it. I personally take about 10s to write the commit message. So such a thing will just make it worse.

I see. didn't think of this one. Note that there is a commit message, one line, and there is an extended description. The extended description, most of the time, is like you said and an LLM cannot do it. But I was wondering if that one line could be auto-generated. If not for every commit, at least for some common changes. But I will take your word that a Language model won't be good.

Of course it is. We use computers to automate things. This is one of them. What is the problem with that?

Thanks. noted.

Aaaaaaand it's slop

Moving on I suppose.

• a lot of default usernames take the form adjective-noun, but I don't know where they get their wordlist

I collected such a word list a while ago, by using my brain to think of words, well, and occasionally removing ones I don't like: https://codeberg.org/trem/fantastic_snail/

The nouns are animal names. They are not the blandest word combinations possible, because I found those difficult to memorize. Instead I went for evoking funny images, sort of anthropomorphizing the animals.

I cannot guarantee that there's not some word combination that could be interpreted offensively, because humans are very creative in offensive language, especially with racism, but I've been regularly generating word combinations for a few years now and remove offensive combinations when I come across them.

It's a Rust library, but the specific word lists are here:

• https://codeberg.org/trem/fantastic_snail/src/branch/main/src/adjectives.rs
• https://codeberg.org/trem/fantastic_snail/src/branch/main/src/nouns.rs

I could also place these lists into the repo as just newline-separated words, if that helps you. Probably makes sense to offer in general...

Ok, let's write an OS in it so people can't complain about devs using AI.

you are both really funny and very talented, where are you from? i wish to know you better

lol I missed that part. This is hilarious. Sadly its proprietary, so we don't know how this was achieved.

Edit: Ah, just realized that's where the name "KillSwitch" comes from. Yeah, I'm a bit slow today.

This reminded me of heroku,which named it's instances (or whatever their called) with random adjective noun this project replicates that: https://github.com/polleverywhere/random_username. Looks like theres other versions out there in other programming languages.