Over the past time I collected some interesting mechanistic interpretability papers - trying to understand the information processing in AI systems.
Since I'm not an expert, I delegated some of the work to AI itself 🙃
Why I find it interesting? Well, the field of "AI neuroscience" is emerging. This, firstly, brings two fields closer together (neuroAI doesn't count). Second, I believe both understanding of AI and the brain can be advanced by testing theories on the systems where everything can be measured and controlled.
Of course, LLMs are far from natural biological nervous systems, but I believe there are fundamental information processing principles that are shared.
💡Some insights: > Activations in RNNs mirror the statistics of their input data > Apparently chunking of computations is present akin to brain's hierarchical processing > "Grandmother cell" are real but sparse, distributed processing is more likely to be happening
Over the past time I collected some interesting mechanistic interpretability papers - trying to understand the information processing in AI systems.
Since I'm not an expert, I delegated some of the work to AI itself 🙃
Why I find it interesting? Well, the field of "AI neuroscience" is emerging. This, firstly, brings two fields closer together (neuroAI doesn't count). Second, I believe both understanding of AI and the brain can be advanced by testing theories on the systems where everything can be measured and controlled.
Of course, LLMs are far from natural biological nervous systems, but I believe there are fundamental information processing principles that are shared.
💡Some insights: > Activations in RNNs mirror the statistics of their input data > Apparently chunking of computations is present akin to brain's hierarchical processing > "Grandmother cell" are real but sparse, distributed processing is more likely to be happening
Have a look, it's fun 🙂
BY the last neural cell
Warning: Undefined variable $i in /var/www/group-telegram/post.php on line 260
The Dow Jones Industrial Average fell 230 points, or 0.7%. Meanwhile, the S&P 500 and the Nasdaq Composite dropped 1.3% and 2.2%, respectively. All three indexes began the day with gains before selling off. Such instructions could actually endanger people — citizens receive air strike warnings via smartphone alerts. "The argument from Telegram is, 'You should trust us because we tell you that we're trustworthy,'" Maréchal said. "It's really in the eye of the beholder whether that's something you want to buy into." In the United States, Telegram's lower public profile has helped it mostly avoid high level scrutiny from Congress, but it has not gone unnoticed. Channels are not fully encrypted, end-to-end. All communications on a Telegram channel can be seen by anyone on the channel and are also visible to Telegram. Telegram may be asked by a government to hand over the communications from a channel. Telegram has a history of standing up to Russian government requests for data, but how comfortable you are relying on that history to predict future behavior is up to you. Because Telegram has this data, it may also be stolen by hackers or leaked by an internal employee.
from us
