Decoding JSON sum types in Go without panicking
I spotted this interesting code stamped into a production plate on this 1990s Mercedes-Benz E300 wagon. The embossed three-digit number codes are the options for the car.
The most interesting part, though, are the hole-punched grid. It has been very difficult to find information on what those codes mean, but I’ve been able to discover that they are 2-of-5 codes, wherein the *spaces* between the punches indicate ones (thick) or zeros (thin), while the *holes* in the last column indicates a check digit, read from top to bottom.
In this case, the six digits appear to be:
11000: 1
11000: 1
00101: 8
10001: 6
10100: 2
01001: 7
Check digit: 01001: 7
The lower-right corner hole seems to be unused.
I don’t know if this exactly right. If you know more information, please reply!
DeepSeek Open Source FlashMLA – MLA Decoding Kernel for Hopper GPUs — https://github.com/deepseek-ai/FlashMLA
#HackerNews #DeepSeek #Open #Source #FlashMLA #MLA #Decoding #Kernel #for #Hopper #GPUs #GitHub #OpenSource #AI
Seriously, Amtrak, why are you using AES encoded data for your train tracking map? (and no API available). 
New short-paper:
"Decoding accuracies as well as ERP amplitudes do not show between-task correlations"
tldr:
=> high decoding accuracy in ERN != high decoding accuracy in e.g. N400
=> strong N170 != strong e.g. P300
This could be great: it could mean that if one ERP-#BCI doesn't work for you, you can simply switch to another ERP!
https://www.biorxiv.org/content/10.1101/2023.05.21.541632v1
Work with Hannes Bonasch, a talented MSc Student in our lab www.s-ccs.de
@eeg @cogsci @cognition #EEG #decoding
