@cogscikid.bsky.social
Kempner Institute research fellow @Harvard interested in scaling up (deep) reinforcement learning theories of human cognition
prev: deepmind, umich, msr
https://cogscikid.com/
prev: deepmind, umich, msr
https://cogscikid.com/
claude code with Opus 4.5
January 22, 2026 at 4:15 AM
claude code with Opus 4.5
codebases almost always have to be rewritten. maybe build initial prototypes with AI, then study the structute and choices and design the final output?
the fast prototyping can be very useful and is almost always wrong in important ways
the fast prototyping can be very useful and is almost always wrong in important ways
December 26, 2025 at 12:57 AM
codebases almost always have to be rewritten. maybe build initial prototypes with AI, then study the structute and choices and design the final output?
the fast prototyping can be very useful and is almost always wrong in important ways
the fast prototyping can be very useful and is almost always wrong in important ways
can you tell us more about why mean field is good for scale? I have a naive mental model that it relies on local averages of populations which feels intuitively wrong? but I'm naive
December 22, 2025 at 5:57 PM
can you tell us more about why mean field is good for scale? I have a naive mental model that it relies on local averages of populations which feels intuitively wrong? but I'm naive
I think it's great as a productivity tool.
I'm talking about getting social support like treating it like a friend or romantic partner (which is shockingly common)
I'm talking about getting social support like treating it like a friend or romantic partner (which is shockingly common)
December 20, 2025 at 11:13 PM
I think it's great as a productivity tool.
I'm talking about getting social support like treating it like a friend or romantic partner (which is shockingly common)
I'm talking about getting social support like treating it like a friend or romantic partner (which is shockingly common)
Reposted
What does this mean? It suggests that architectural inductive biases alone can get us surprisingly far in explaining the image representations of the ventral stream. See a great commentary by @binxuwang.bsky.social wang.bsky.social and Carlos Ponce. www.nature.com/articles/s42...
Structure as an inductive bias for brain–model alignment - Nature Machine Intelligence
Even before training, convolutional neural networks may reflect the brain’s visual processing principles. A study now shows how structure alone can help to explain the alignment between brains and mod...
www.nature.com
December 15, 2025 at 6:45 PM
What does this mean? It suggests that architectural inductive biases alone can get us surprisingly far in explaining the image representations of the ventral stream. See a great commentary by @binxuwang.bsky.social wang.bsky.social and Carlos Ponce. www.nature.com/articles/s42...