GL26

5 ms·

1.
by GL26 3mo ago
Based of Tom Blomfield's talk about self improving companies with AI, he talks about loops present in companies that make AI agents self improve. He talks about loops that look like this : - Data acquisition / feeding mechanism -
2.
by GL26 3mo ago
Get ready for the great revolt
3.
by GL26 3mo ago
The POC was made to classify materials, we couldn't get our hands on scientific grade materials that would allow to test our hypothesis, so we falled back to just identifying materials. We even went to identifying multilayer materials.
4.
by GL26 3mo ago
Yes, it is half baked, we stopped because the quantity of data needed to continue was too expensive. Concerning the materials that were in the confusion matrix that you saw, it was only non asbestos materials (plexiglas doesn't ever co
5.
by GL26 3mo ago
Your comment still has a lot of value, I didn't detail it in the article but this radar is usually used for autonomous vehicles, or person tracking.
6.
by GL26 3mo ago
Hi, thank you so much for your comment, the research was cool to do, but the market isn't there, and this is not something people really want (to retake YC's wording). You can have all the tech you want, but at the end of the day,
7.
by GL26 3mo ago
small note : the website wasn't coded to handle such a trafic, I had to inject a few netlify credits to keep the website alive. The videos are now linked on youtube (they are initially unlisted). Got actually ddosed by HN (good problem
8.
by GL26 3mo ago
you could definetly equip you wall-e with this sensor, but honestly a camera with yoloV26 + fine tuning for trash recognition would be enough
9.
by GL26 3mo ago
exact, if I was carrying a PhD (which I am not). I think I would have wanted to go out further and get more data to make this device pass regulatory tests. What was the status quo we were solving for ? An asbestos analysis costs around 60€
10.
by GL26 3mo ago
overtech for a problem that had a solution (asbestos sensing is pretty painful in Europe), but anyways the market was shrinking, and the TAM was totally not VC backable. Tested it out with : wood, copper, alumnium, paper (the book you saw),
11.
by GL26 3mo ago
the FMCW tech makes it impossible to have a resolution inferior to 2.5 cm, (so if two layers are appart, you can't physically tell them appart using physical classical modeling techniques with DSP). However, you can use AI to enhance t
12.
by GL26 3mo ago
hahaha ! oops didn't mute the video, would blast trap music when I was alone in the lab x)
13.
by GL26 3mo ago
thanks :) !!
14.
by GL26 3mo ago
thank you so much for your feedback, it was hard to admit defeat, but at the end looking back at what I built, the parts where I learnt about RF, and just struggled, refactoring the code for the sim (thank god cc is not good enough to under
15.
by GL26 3mo ago
just fixed it, hope it works
16.
by GL26 3mo ago
My netlify crashed fixing the website rn
17.
by GL26 3mo ago
How many tx and rx antennas did you have ? (I don’t know if it was clear, my stack was 57-64 GHz, 2TX , 3RX)
18.
I built a mmWave material classification radar (2025)(gauthier-lechevalier.com)
203 points by GL26 3mo ago|60 comments
19.
by GL26 3mo ago
Feels like identity control is not such a bad thing (prevents bots, identity thefts, cyberbullying, ect...). The questions is not "if it happens", its more a "who controls it?". If an NGO controls it, or a company that h
20.
by GL26 3mo ago
The hack "fiu this is you from the future" is genuinely funny. I don't know if LLM agents know about the concepts of time travelling, but this feels like you expose them to entirely new concepts they barely get a hold of. (By
21.
by GL26 3mo ago
This is very satisfying to play. Saw this game on steam that seems close in the « Rogue Like Chess » concept : https://store.steampowered.com/app/3509230/Gambonanza/
22.
by GL26 3mo ago
This would be huge if it was « rogue like », where you could buy new more performant components that allow you to reach further into the game. The game makes you naturally lose, and there are milestones that you can reach (bosses, or loot t
23.
by GL26 3mo ago
PyTorch already does a huge job at keeping DL not verbose. It captures the complexity to allow you to make a whole GPT model with it, running on your GPU, and keep the code readable. Yes, there are a lot of breaking points, but at some poin
24.
by GL26 3mo ago
Very instructive on how a computer rack works. I cought only later on that it was Oxide specific, but the design decisions seem so obvious to me that they look like industry standards. What about firmware architecture ? How do you design fo
25.
by GL26 3mo ago
Problem with thinking you can replace your employees with AI, this is not the case. This is like thinking you could replace your NASA engineers with IBM computers in the 60s. The AI revolution changes drastically the way people work, and em
26.
by GL26 3mo ago
At last, they got the FOMO from GLM-5.2 taking their place. Can't wait to try and try to bypass the guardrails to use it for cybersec capabilities
27.
by GL26 3mo ago
Embedded engineer here, there are a lot of concepts of "thread priorities" that are great to steal from RTOS, and connectivity features that are made to ensure you never ever break the minimal "software" that is required
28.
by GL26 3mo ago
insane ! I don't know if it's possible but it would be huge if we had access to the localisation of the trends
29.
by GL26 3mo ago
Spoiler alert : Singapore won the race years ago. Cameras everywhere, and mostly : the singaporian civilian population is educated to surveil peers so that they don't commit incivilities. Here is an article about it : https:/&#x2
30.
by GL26 3mo ago
if someone has any tutorial on how to run GLM-5.2 from a Rasberry Pi 5 (AI hat), I want it !

More ›