roger_

I personally don’t care who has access to my health data, but I understand those who might.

Either way, I’m excited for some actual innovation in the personal health field. Apple Health is more about aggregating data than actually producing actionable insights. 23andme was mostly useless.

Today I have a ChatGPT project with my health history as a system prompt and it’s been very helpful. Recently I snapped a photo of an obscure instrument screen after taking a test and was able to get more useful information than what my doctor eventually provided (“nothing to worry about”, etc.) ChatGPT was able to reference papers and do data analysis which was pretty amazing, right from my phone (e.g fitting my data to a model from a paper and spitting out a plot).

Is there a mainstream distro that disregards all the legacy cruft? Gobo, but that’s not really mainstream.

Mac OS?

Last time I look into SoTA Bayesian deep learning, Bayesian output layers seems the most promising and practical. Is that still the case?

I’m sure you can point Claude at that page and have it make the necessary changes to pass.

Pretty shocked that Xiaomi publishes the protocol: https://iot.mi.com/new/doc/accesses/direct-access/embedded-d...

My Thingino camera seems to have some RTSP issues, which is a shame because I’d like to use it on more devices.

2.4 GHz sucks though. Wish my mesh allowed me to use multiple 2.4 GHz channels concurrently or per node.

I have around 6 cameras, mostly 1080p, and about 8 GB RAM and 3 cores on the VM (plus Coral USB and Intel VAAPI). CPU usage is about 30 - 70% depending on how much activity there is. I also have other VMs on the machine running container services and misc stuff.

There are some camera stability issues which are probably WiFi related (2.4 GHz is overloaded) and Frigate also has its own issues (e.g. with detecting static objects as moving) but generally I’m happy with it. If I optimize my setup some more I could probably get it to a < 50% utilization.

IncusOS looks pretty interesting. It’s a new immutable distribution designed for Incus.

When it matures I’ll look into switching from Proxmox.

Yup my primary Windows machine is a VM and after passing through all the relevant peripherals (GPU, USB) it’s pretty seamless and you’d never know.

Cool part is I needed a more powerful Linux shell than my regular servers (NUCs, etc.) for a one off project, so I spun up a VM on it and instantly had more than enough compute.

I have Frigate and a Coral USB running happily in a VM on an N97. GPU pass through is slightly annoying (need to use a custom ROM from here: https://github.com/LongQT-sea/intel-igpu-passthru). I think SRIOV works but haven’t tried. And Coral only works in USB3 mode if you pass the whole PCIe controller.

I noticed your feature vector is large and you don't use ML. What's the final statistic that you threshold?

Actually I misspoke. I previously used STA/AP mode (and two ESP32s) but I switched to something close to what you describe. I filter the pings to only get the ones targeting a specific MAC (in promiscuous mode). This way I get only specific CSI packets and they're perfectly periodic at whatever rate I want.

Sounds like your MVS approach is a sliding window variance of the cross channel variance, with some adaptive thresholding. My pre-processing has generally been an EWMA de-meaning filter followed by some type of dimensionality reduction and feature extraction (kernel or hand-crafted, like raw moments), which I think fits into your overall architecture.

I'll have to look more closely at your work, thanks for sharing!

Been working on this very idea casually for couple years with ESP-IDF and I could never get the statistical signal processing just right (by my definition). Things I've tried: adapting filtering (LMS, Kalman), kernel methods (NEWMA, MMD), detectors (CUSUM, GLR), dimensionality reduction (random projection, online PCA), whitening, etc.

I use a single ESP32 in STA/AP mode which sniffs ACK packets with a specific destination mac, which come from any server on my WiFi network (uses a special sniffing mode IIRC). This way I can receive regular CSI packets originating from a fixed location and doesn't need another device running.

I'll have to look at this code, maybe I just overlooked the obvious or my requirements were too high!

I kinda wish there was a modern *suite* of improved tools, developed by one team with consistent designs (parameters, colors, tables, etc.)

Copilot in VSCode seems to do something similar in the form of todo lists.

Agreed.

The nushell homepage, by comparison, immediately conveys the benefits of that project.

Interesting, can this be applied to regression?

ESPHome is good for that.

I’m surprised engineered prompts are still needed. Most LLMs seem to do a pretty good job nowadays if you just say what you want.

I feel the same way about Beetlejuice Beetlejuice and Napoleon Dynamite, but just wanted to add that Michael Keaton’s portrayal was still brilliant and I’m happy he could still pull it off.

The plot and all the non-Beetlejuice scenes were a waste of time.

“Bundled with alternatives to over $5,000 of professional software”

That sounds spammy and misleading to me. I’m assuming they’re just including open source alternatives and assuming the same value as commercial offerings.

Do the detectorists get a certificate or something for their honesty?

Great, looking forward to it!

I’ve used RomCenter and other tools in the past and I’d love to see similar, comprehensive functionality in something more modern.

Been tracking this for a while and it really needs support for renaming and validating ROMs (e.g. with .dat files) to be called a “ROM manager”.

Been following wesselb on GitHub for a while, great to see his work getting more attention!

Won’t the changes eventually be added to ollama? I thought it was based on llama.cpp

Never got how mamba models work in multiple dimensions and non-causally.

RIP. Still waiting for someone to complete the final exam from Real Genius: https://imgur.com/a/real-genius-exam-questions-SjBhm

Thanks! Does codemcp support having the server on another machine? Maybe communicate over ssh?

I’m interested in using MCP for pair programming.

What tool can I use to point it to a directory and give Claude access to the code? I don’t want to have to write my own server.