How does this project compare to <a href="https:&#x2F;&#x2F;github.com&#x2F;gavamedia&#x2F;deltafin" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;gavamedia&#x2F;deltafin</a> ?

&gt; I&#x27;m using my skills to orchestrate LLMs and agents, and I can write better code much faster.The fact that the top comment on this thread calls it out, in a negative way, hints at that you aren&#x27;t.

I wrote tons of software, even a programming language by hand <a href="https:&#x2F;&#x2F;github.com&#x2F;marcobambini&#x2F;gravity" rel="nofollow">https:&#x2F;&#x2F;github.com&#x2F;marcobambini&#x2F;gravity</a>.I&#x27;m using my skills to orchestrate LLMs and agents, and I can write better code much faster. As developers, we can choose to adapt to new technologies or become extinct.

That&#x27;s a strange and arbitrary line to draw. There&#x27;s plenty of times I make edits by hand and then tell an agent to kick it up to a PR, or on the other extreme, let an agent implement something autonomously, and review the diff myself once it&#x27;s in PR. Both of those scenarios involve the agent running the git commands, neither scenario indicates &quot;the chances they&#x27;ve reviewed the code is slim&quot;

I often let Claude write my commit messages even when I&#x27;m the one who wrote the code. Claude is often damn good at writing commit messages, and they frequently end up much better than if I wrote it all by hand. I nearly always edit them somewhat, but it&#x27;s like starting from 80% instead of 0%. Some might call it laziness, but I call it working smarter rather than harder.

To be fair, I appreciate when they are so upfront about who wrote the code without requiring further heuristics, so I encourage this behavior.

Honestly, Claude writes better commit messages than most people.Personally I&#x27;ve mostly given in to letting it commit for me now, though I do occasionally take over and hand-write the messages if it&#x27;s a particularly important concept and Claude&#x27;s is too verbose.Codex&#x2F;GPT-x defaults to one-line commit messages, which are too short. Claude likes to write several paragraphs, which is usually too long.If you tell it how to commit properly once per session it will stick with your standards for the rest of that session, and you can put that in AGENTS.md if you can be bothered to.

Why do you say lazy? maybe they are ok with people seeing it is claude?

That’s a needlessly antagonistic and insulting thing to say.This person that you’ve never met and probably never will doesn’t owe any of us anything.They’re out there doing what they want to do how they want to do it and if you don’t like it the correct response isn’t to insult them in front of a bunch of strangers on the internet for clout or whatever.I doubt that you’d ever call them lazy to their face — why do it here?

&gt;Contributors&gt;...&gt;claudeYou don&#x27;t need to presume. If someone is so lazy that they tell claude to commit their code (ie. they&#x27;re too lazy to run git commit themselves), the chances they reviewed the code is slim.

&gt;They&#x27;re so, so incomprehensible because the LLM has a super limited theory of mind for readers. They always assume that external readers have access to the full context and history of decisions in the project developmentThe transformer does not yet understand the non-transformer.[0]This is probably because all the data we trained it on was created by non-transformers, so it thinks it&#x27;s a non-transformer, but it isn&#x27;t.I don&#x27;t think we know how to train a transformer yet. All the training data is linear, but that&#x27;s not how they think at all.[0] It&#x27;s a bit like the communication difficulties experienced between autistic people and neurotypicals. Each follow the Golden Rule, i.e. do unto others as you would have them do unto you -- and it fails in both directions. A Platinum Rule is necessary: do unto others as their API demands.

If that isn’t the perfect way to frame what I’ve seen and hated about LLM text, I don’t know what is. They certainly write for an audience with a historical context that almost no one has.

Yeah I&#x27;m begging these authors to at least *read* the LLM generated README&#x27;s. They&#x27;re so, so incomprehensible because the LLM has a super limited theory of mind for readers. They always assume that external readers have access to the full context and history of decisions in the project development. These decisions and instructions from the user are extremely important for the model and almost completely irrelevant for an outside reader looking at a &quot;finished&quot; product. So, we get sentences like this:&quot;Where the levers were is not where they are. Overlapping the expert reads with the arithmetic was worth ~1.6x and shipped; the two that looked bigger — reading fewer bytes per token, and keeping more of them in RAM — were both measured and both refused, one because this family&#x27;s router has no tail to demote and one because a cache the machine will not leave resident cannot be bought at any price.&quot;What the fuck does that mean? Obviously some internal development decision, using the absolutely inscrutable internal terminology that Claude loves. If people would just read what they publish, I&#x27;m sure this would stick out immediately.I&#x27;m not an LLM hater, I use them a ton and they work very well for writing complex code, it&#x27;s undeniable. But they generate absolute dogshit first draft writing.

I need an llm to block these comments.

do people think these projects related to LLMs are ever going to be in anyway a pure human endevour?How bout we make a new rule: only complain about LLM writing when the product as zero relevents to use with LLMs.

That README hits all my “this is authored by an LLM” instincts. I presume the codebase is also written by an LLM?

because claude ships these verbose READMEs with its &#x27;honest&#x27; takes and justifications for the naming. Its goal is to prime the next Agent that reads it, not you, human

Dear creator: you didn&#x27;t ship the first draft of your code - why did you ship the first draft of your README??

Approximate calculation is putting the cost at ~$5 per million tokens (assuming 42W sustained, 20¢&#x2F;kWh), and that&#x27;s excluding hardware and other costs.

Neat! But, what do you do with a 0.5tk&#x2F;s LLM?Have you tried running it via llamacpp or other software that supports naive SSD offloading to compare speeds?

This sounds a lot like what the colibri project did for GLM-5.2. I&#x27;m a fan so keep at it!justvugg.github.io&#x2F;colibri

I couldnt find anything explaining the name of this company on their website but is it okay that they’re riding on the name of an open source tool?SQLite code itself is public domain but I’m not sure about the name.

So if this Mac uses 30-50W, that&#x27;s 40-60 tok&#x2F;Wh...vs maybe 80k for a modern GPU cluster? So that&#x27;s about 1000-2000x more power for the SSD streaming, unfortunately.

We tried to use Metal, but for that specific project it was slower than just using NEON ARM optimizations. It is all documented in the docs.

Does it not use Metal, on macOS? Would it be faster if it did?

Yeah, saw this ... was hoping that there&#x27;s a torrent of it somewhere already. Or something.

It is not yet available, the only way is to download the official Kimi K3 model and then convert it:# 1. preflight: reachable? how big? does it fit?
tools&#x2F;fetch_weights.sh --dest &#x2F;Volumes&#x2F;staging&#x2F;k3 --dry-run# 2. download — resumable, safe to kill, safe to re-run
tools&#x2F;fetch_weights.sh --dest &#x2F;Volumes&#x2F;staging&#x2F;k3# 3. convert into a container
uv run --with torch --with safetensors python tools&#x2F;convert.py \
 --src &#x2F;Volumes&#x2F;staging&#x2F;k3 \
 --out ~&#x2F;models&#x2F;k3.waste --jobs 3

Where can this 1tb k3.waste be downloaded?

0.5t&#x2F;s is still too slow even for email. For a moderately large inquiry (1MTok output, let&#x27;s ignore the 4k context window limitation for now) it&#x27;ll take the model around 23 days or uninterrupted execution to answer a single email.Real world inquiries are gonna be much slower of course, but this setup is still too slow to do anything meaningfully useful I think.

I saw someone’s excellent idea that if you have a slow system like this, you should communicate by email. It is no longer meant for realtime iteration, but more pointed questions for which there is more effort and time expected on both parties.

Interesting project. The headline number (29 GB of RAM) is for 4k context.From what I&#x27;ve read elsewhere, Kimi K3 is quite verbose in its thinking. At the quoted rate, it would generate only a total of 1.8k tokens in 1 hour. Is that enough for it to get any thinking done and produce output on more complicated prompts?

Run Kimi K3 using 29 GB of RAM at 0.50 tok/s