Having read plenty of Dr. Oliver Sacks (e.g: Man Who Mistook Wife for Hat), there has almost-certainly been a savant ("idiot," or not) with this actual ability – and nobody knew how it worked (it just did), similarly to the documents Sacks-patient that immediately knew how many toothpicks were on any table (in the hundreds-to-thousands)... but could otherwise not talk.
----
To explain this with a DCC metaphor:
NEW ACHIEVEMENT: All prime numbers now glow bright orange, including entire mobs (if they contain prime number of living members). This includes those cute little toes of yours, Carl. Don't question this, it just is.
----
My favorite Sacks patients were the non-verbal aphasics laughing at Nixon's public TV appearances (because they were experts at telling if people were lying [by laughing at non-verbal cues]), otherwise disconnected from society.
Yes... I wonder if this is also prone to hallucination? A while (more than a year) ago I told Copilot to sort a list of integers. First, it gave me the code to sort it. I told it "no, sort the list yourself and give me the result". Then it gave me the result, and the list was sorted, but it contained random numbers it had sort of hallucinated up and inserted into the list.
Is there something special about yolov8 over later models (9-12)? It seems most of the research and working examples default to v8 despite it being 3 years old. Or just because it is what fits on this hardware?
Mainly because YOLOv8 is well-supported by the Rockchip/RKNN toolchain.
The goal here was an end-to-end RK3588S pipeline rather than comparing detector families: training/export, ONNX graph fixing, INT8 RKNN conversion, C++ postprocessing, and runtime inference across the 3 NPU cores. YOLOv8 has known-good export paths and Rockchip examples, so it was the most practical baseline.
Newer YOLO versions may be possible, but usually require more work around RKNN export compatibility.
I didn’t know how electrolysis really worked, and probably still don’t but I know a little more, and these cheap ion filtering membranes are absolutely wild, and turning the graphite fire blanket into a super high surface area electrode is super interesting. I remember when he re-made the “lost” aerogel-like substance a while back and wonder if we can make graphite aerogel-likes using similar processes somehow for super high surface area electrodes. Sometimes I think I’m more interested in practical materials science than software engineering, but at 39 years old I am probably past the point of going to university to actually study. Of course, unless we all get UBI quickly because of Claude 8.5 taking everyone’s jobs.
> at 39 years old I am probably past the point of going to university to actually study.
That largely depends on your financial situation. If you have a strong technical background and you've already secured your retirement you could certainly do a masters or phd. But if you aren't financially secure then yeah, accepting (somewhat worse than) minimum wage for the next 5 years followed by a high degree of uncertainty sounds like a really bad idea.
That said materials science is something of a bastard child of inorganic chemistry, applied physics, and engineering. The theory side of it can be absolutely brutal. Before embarking on an adventure I'd suggest looking over the coursework for physical chemistry to see if you can handle the quantum mechanics stuff.
The problems I was interested in led me to the techniques in the field. I guess the idea was data science, but might need to pivot into LLM or AI engineer.
reply