bitexploder

7 ms·

61.
by bitexploder 26d ago
I have basically polluted my context and agent files, agent.md, claude.md, gemini.md with really strong guidance about using plain language. I even include small, mini design specifications, examples of plain engineering language, only usin
62.
by bitexploder 27d ago
I decided not to get used to its communication style. It encourages it to invent terminology and drift away from simple and proper engineering in my opinion. Also, it is pretty simple to change as long as you’re not using the Claude code CL
63.
by bitexploder 27d ago
It seems a little excessive to use another LLM. With OMP I basically created an ephemeral prompt stack all of my agent files. It walks up the directory tree looking for any Gemini.md, Agents.md or Claude.md files. And it puts those at the v
64.
by bitexploder 1mo ago
I have been using Muse Glimmer for a few days instead of A3B. It gets the job done quicker than A3B despite being several times slower.
65.
by bitexploder 1mo ago
Yeah, but be fair. Working with small models is a different ball game. Not all the batteries come included :)
66.
by bitexploder 1mo ago
I had to fix this on 35B A3B -- I have a proxy that just shuts it down if it gets to 2K thinking tokens and injects something like "We have thought enough, let's begin working." and it almost always finishes the turn then. It
67.
by bitexploder 1mo ago
GLP-1 does a lot more than just stomach signaling. It shrinks visceral fat, it's been shown to clear liver fat, and it actually increases mitochondrial efficiency similar to if you were to do endurance work/cardio. There's a
68.
by bitexploder 1mo ago
They controlled for BMI at least, so they knew that was going to be a common question. BMI is a pretty good proxy for that, and they still said that the majority of the effect came from the compound and not BMI.
69.
by bitexploder 1mo ago
That is exactly how this model has worked for me so far. Muse on a one-shot task will get to 80%. And if you even nudge it and say, "Hey, finish up," or "Review the syntax," boom, it's done. And I'm getting 20
70.
by bitexploder 1mo ago
Have you tried Muse 30B yet? I have been impressed with it. I have Qwen 3.8 27B hammering away right now against Muse. And Muse is doing a little bit better.
71.
by bitexploder 1mo ago
Flash models and Gemini make more sense when you consider Gemini Enterprise and Workspace. Oh HN we generally care a lot about writing software. However, until Fable, Gemini 3.1 Pro was my default for doing any sort of discussion outside of
72.
by bitexploder 1mo ago
I would also think about combining the descriptions of photographs with some sort of RAG to essentially help shoot you into the reference references easier.
73.
by bitexploder 1mo ago
I ran a 9B over my like 100k photo library — it was very good at it. And extracting any text. All local.
74.
by bitexploder 1mo ago
Pi and a similar set of tools is also likely similar to the harness these models are trained on. More complex harnesses burn reasoning tokens on these small models and in my benchmarking don't seem to be able to beat Pi ever. Usually i
75.
by bitexploder 1mo ago
You really have to get the models to end their thinking. Almost any commercial model serving has safe guards like this to tune how much they think.
76.
by bitexploder 1mo ago
I have a custom A3B proxy that caps its thinking off. It is a known issue with the model that Qwen themselves documented but is almost never addressed in any harnesses. I also patched up a few other known bugs in the proxy. I highly recomme
77.
by bitexploder 1mo ago
Usually something like dsv4 running them. Depends in the horizon. It’s kind of an overnight thing for the project. Use matt pocock skills or similar. Build good spec. Iterate on it with a big model or your brain. Break it down into pieces a
78.
by bitexploder 1mo ago
It is the only possible way out. The level of tech debt can only be paid by more agents. But seriously, it is an actual bet you can make and it isn’t crazy sounding to me. Build a world where you have an edge with future models and can surv
79.
by bitexploder 1mo ago
Work: Security engineering focused on securing authnz at hyperscale in our new agentic reality. What interests me? Agentic engineering, vulnerability discovery at massive scale. Vulnerability remediation at massive scale. Bounding agent non
80.
by bitexploder 1mo ago
Have better agents rewrite it in the future.
81.
by bitexploder 1mo ago
OpenRouter and Pi here now.
82.
by bitexploder 1mo ago
I find 35B A3B viable as well, but your harness and runtime really matters to get tool calling and such dialed in. In fact, I would encourage you to experiment with it some as I find I get more reliable output from 35B A3B, though 27B is st
83.
by bitexploder 2mo ago
Oh… definitely not enough text. LLMs don’t work like that. Hmm.
84.
by bitexploder 2mo ago
You could build a tool like that and on a MBP with 48+ GB of RAM all of that can happen locally in terms of keeping your content locally. I am sort of an outliner and planner and I heavily use AI when writing documents for work. I don'
85.
by bitexploder 2mo ago
I am going to see what I can get out of this streaming from SSD vs. Qwen 27B. Laguna is a seriously impressive model. I already have a lot of success with 27B A3B.
86.
by bitexploder 2mo ago
AOL existed…
87.
by bitexploder 2mo ago
Same story as “Oh noes, Mythos too powerful”.
88.
by bitexploder 2mo ago
That is what I have learned over the last year. I still only have so much mental bandwidth. I can definitely do more now, but I have become relatively disciplined about what I nudge along each week with agents and it's okay if it doesn
89.
by bitexploder 2mo ago
We now have more work, not less. Engineers can finally address security and tech debt! Surely businesses will see the value in that. Right?
90.
by bitexploder 2mo ago
You know what else increases the odds you can’t brute force it? Not having one :) Seriously though, the best password boxes are one that have an entropy meter and check that your password has never been in a breach.

More ›