I'm liking all of the remixes.

We built a combined books/video version as well in our app library.

https://medium.com/@vektormemory/vektor-2-0-4-hacker-news-bo...

Great idea, I do a lot of work testing small LLM's in this space and agree if you add some guardrails and train a small embedder, you won't need these large models, as your info is static, you can ship it all as one.

Or try these with your trained embedder: https://ollama.com/library/minicpm-v4.6 https://ollama.com/library/bonsai-2

If only there was a company that didn't use ads, protected people's privacy, didn't sell their data, and had local data sovereignty?

Full article with desk routing benchmarks

We have to test many different api's and models as we have over 600+ in our harness now, Sonnet 5.5 via cloud code provides the highest quality results.

Over the last few years we have added mcp tools, memory graph, md files and skills, automated everything to the point that on desktop the system runs itself, besides proceed/approved/explain.

We have 100's of PRs/jobs/projects and Claude builds down the list automated and only asks a few questions point form if needed.

Very impressive now, compared to 6 months ago.

You can view it if interested: www.vekmem.com

Very impressive indeed https://imgur.com/a/xbwdkWc

Full unpaywalled article

A full unpaywalled article is provided.

And yet, even at the price of free, isn’t worth reading.

Please pull the little orange mushroom cap out of your mouth before re-posting.

Those close x's are exactly how the govt would make them, peak design...

80% Sonnet 5.5, Opus 5.5 to finish the last 20% cleaning up the fine details.

Gemini 3.1 pro as a nutty professor, researcher, and design verifier.

My custom orchestrator for all local LLM work: https://vektormemory.com/vektor

Can someone remove the extra LLM and just have an embedder do the classifier work?

It's turning into pimp my llm...

I simply use an embedding model followed by a simple MLP, and that’s enough to solve many text classification problems.

My little pony ?

Multi layer perceptron. A tiny neural net.

Boosted trees over embeddings work surprisingly well too.

Wait until you hear about support vector machines!

Its funny - I was going to leave a similar comment - and I have, earlier, on a different thread. If people need fast classification, on a fairly scoped problem, it is very fruitful to start with an off-the-shelf embedding model like ModernBERT (which Laya uses) and stick a classifier in front - like a Support Vector Machine (SVM). For starters just tune the SVM, you don't even have to fine-tune the embedder - often it works very well, esp. given the compute needed. Plus you can get reliable confidence scores and generate explanations if you want them (using something like SHAP).

If you have an easy problem that can be solved by a classifier from pre LLM era, then sure, go ahead. But we have LLMs now and we can use those to expensively and slowly classify harder problems using general purpose models, without needing to spend a huge amount of time fine tuning a model for the specific task. Jev offers to do the same fast and cheap.

For clarity: no, a 0.8B model is not gonna do that.

I had the same thought: What were we all missing these past three years? And I think it's just that while everyone was staring at LLMs, the hardware has become a lot better and we now have 0.8B models that are extremely well-trained, so that allows us to look at the "old tech", such as classifiers, in a new light. It's also interesting because it decentralizes innovation: You don't need billions of hardware to do interesting experiments in this space.