A friend of mine asked me, "How can we give models their own sense of taste?" Furthermore, he remarked that typically we use humans as...

I'll admit , "antipsychotic" is a bit of a scary word. The second half, "psychotic," refers to some of the most extreme and frightening experiences a human can have. I think the word evokes imagery of psychiatric wards, confusion, strait jackets, possibly violence, and overall distress. A treatment designed to suppress such a serious condition might suggest being powerful enough to match the strength of the condition, thus giving antipsychotics their own stigma—especially if

In another post , I shared some thoughts on why the physical boundaries of You  might be difficult to make solid on the basis that...

I'd like to make two distinctions. An Instantaneous You  and a Distributional You . The Instantaneous You  represents a singular point...

Something I am curious about is the role of other people, or other stimuli that prompt awareness of self, in influencing the formation of...

The most common generative models in deep learning include GANs, diffusion models, and autoregressive models. What they all have in common is that they are all methods of both modeling and sampling from p(x), a data distribution of interest, just different methods of decomposing the problem and relying on varying assumptions. Though in the end, all of them can do both: Create convincing samples that somewhat blend in with other real data samples. Give an estimate of the proba

We appear to experience life through the point of view of a discretely singular, egocentric entity, distinct from everything else. Pride...

There was an interesting quote from Janus's article on simulators. "GPT" is not the text which [it] writes itself Furthermore, they...

From an evolutionary perspective, it's clear that humans have made it to the top of the food chain, but what did it take for us to reach...

In our pursuit to perfect neural networks, we often look to how humans learn for reference, which has had varying degrees of success....

Many facets of our universe seem to gravitate towards unstable, chaotic equilibria. One such example is criticality, a threshold where a...

In another post, I wrote about parameterless (boneless) attention as a means of mixing information across datapoints weighted by their...

I felt compelled to make this post after seeing yet another reinforcement learning paper for diffusion models that does spectacularly in...

We are racing toward an uncertain future that draws nearer every day. We have a rough idea of what the future could look like. I'm inclined to think some of our guesses and fictional depictions may hold some weight. Historically, we've been good at that. Possibly now more than ever, we understand ourselves, as humans, to know the things we will seek and the trend we are on. Meanwhile, in our sciences, theories and understandings have outpaced implementation, extending our vis

Something I've been thinking about a lot lately is how humans handle noisy recognition. Maybe you recognize the image above, if not you...

Once upon a time, I was set on becoming a psychiatrist. Throughout life, I've spent probably a near unhealthy amount of time thinking...

The Distillery - a look under the hood of vibe coding Introduction Vibe coding may be one of the best and worst things 2025 has had to...

I’ve seen a lot of convoluted tutorials on attention but nothing really made it click for me more as understanding as mixing a projected...

One conviction I hold very strongly is that "special" people are possibly much more common than we may be lead to believe. While sure, we are all special in our own way, in this post, "special" refers to having seemingly super-human skills that lead high social standing, being seen as an expert of a field, and generating significant economic, cultural, or scientific impact. Not necessarily the acts you see on America's Got Talent, but the skills that make us think of child pr

Presently, most optimizers used in deep learning do not explicitly accommodate their updates with respect to the expected range of...