zlacker

[parent] [thread] 4 comments
1. btown+(OP)[view] [source] 2025-10-22 16:17:32
In all seriousness, if AI models can handle quantization, they can handle some flipped bits from time to time! There are probably some fascinating papers to be written around how to choose which layers in an LLM architecture could benefit more than others from redundant computation in a high-radiation environment.
replies(1): >>kibwen+G5
2. kibwen+G5[view] [source] 2025-10-22 16:44:03
>>btown+(OP)
Brilliant, to turn up the model temperature we just hinge open the shielding. I call dibs on the patent!
replies(1): >>lawles+Uq
◧◩
3. lawles+Uq[view] [source] [discussion] 2025-10-22 18:18:36
>>kibwen+G5
Ok, has anyone patented chips with radioactive source glued to them? For "true" randomness.

If it not i want dibs on it.

replies(1): >>DontBr+Tv
◧◩◪
4. DontBr+Tv[view] [source] [discussion] 2025-10-22 18:42:20
>>lawles+Uq
https://en.wikipedia.org/wiki/Hardware_random_number_generat...

> and even the nuclear decay (due to practical considerations the latter, as well as the atmospheric noise, is not viable except for fairly restricted applications or online distribution services)

replies(1): >>notaha+PV
◧◩◪◨
5. notaha+PV[view] [source] [discussion] 2025-10-22 20:52:37
>>DontBr+Tv
yeah, I think the space weather experts would have fun statistically analysing the single-event-upset RNG :)
[go to top]