All Notes: Nonlinear Function

All Notes

limiting belief

Modified: October 29, 2021.

Tagged with: #meditation#psychology

limits on individual impact

I want to change the world. What does that mean? Suppose I create a billion-dollar company. That's an enormous amount of value. It's many…

Modified: February 14, 2021.

Tagged with:

linear attention

tags: created: 2023-12-07 modified: 2023-12-07 References: https://arxiv.org/abs/2006.16236 The usual [ transformer ] [ attention…

Modified: December 07, 2023.

Tagged with:

linear time-invariant

A linear time-invariant system is one where the dependence of the output on the input is: linear: an input produces an output , and…

Modified: December 05, 2023.

Tagged with:

lipid bilayer

Modified: .

Tagged with: No Tags

listening inbox

podcasts and audiobooks Alan Watts 'out of your mind' lectures

Modified: February 15, 2022.

Tagged with:

live the life you want to have

I've heard it said, and it's been ringing true to me, that the thing to do is live the life you want to have now , not plan to spend years…

Modified: February 12, 2022.

Tagged with: #life-advice

living on autopilot

Some things are genuinely hard to do. But many others I don't do just out of laziness, or maybe lack of [ agency ]. I know that they're…

Modified: September 12, 2021.

Tagged with: #personal#mental-health

logical induction

Sources: https://www.lesswrong.com/posts/y5GftLezdozEHdXkL/an-intuitive-guide-to-garrabrant-induction https://www.lesswrong.com/posts/jtMXj…

Modified: April 02, 2025.

Tagged with:

lone pair

A atomic orbital with two electrons both attached to the same atom. In contrast to a bond, where each atom contributes one electron. Lone…

Modified: May 14, 2021.

Tagged with: #chemistry

long-term context in Transformers

Notes on https://www.pragmatic.ml/a-survey-of-methods-for-incorporating-long-term-context/ 'Standard' transformers have O(n**2) complexity…

Modified: March 21, 2020.

Tagged with:

long-term potentiation

Modified: .

Tagged with: No Tags

looking under the lamppost

There's a tendency to focus on things that we have the (conceptual/mathematical/societal) tools to understand, even when we know this is…

Modified: February 10, 2022.

Tagged with: #how-to-think#research

loose veganism

How do I justify being only 'mostly' [ vegetarian ]? I know that cows and chickens are abused to produce milk and eggs. Why is avoiding…

Modified: February 20, 2021.

Tagged with: #morality

love

Modified: .

Tagged with: No Tags

love is value alignment

What does it mean to [ love ] someone? Of course this question has as many answers as there are people, and probably more. But here's one…

Modified: November 28, 2023.

Tagged with: #ai#alignment#relationships

love is value alignment.sync-conflict-20240609-185651-QL2G4SK

What does it mean to love someone? Of course this question has as many answers as there are people, and probably more. But here's one view…

Modified: November 28, 2023.

Tagged with: #ai#alignment#relationships

love-positive

Modified: November 28, 2020.

Tagged with:

loving-kindness

see also [ metta ]

Modified: April 10, 2024.

Tagged with:

lustful curiosity

I saw this phrase on Twitter somewhere and it really resonates as a description of the ideal approach to science. There is no real…

Modified: May 16, 2022.

Tagged with: #how-to-think

macrostate

A macrostate in statistical mechanics is a collection of base-level states; equivalently, a subset of [ phase space ]. It's what you see…

Modified: April 13, 2022.

Tagged with: #physics

magical display

Modified: .

Tagged with: No Tags

mahamudra

Mahamudra means the 'great seal' or 'great gesture'. We take and [ hold the view ] that each event arising in [ awareness ] --- every sight…

Modified: October 09, 2022.

Tagged with: #meditation

manager

Managing for high-variance / creative work versus low-variance consistent work: https://blog.sbensu.com/posts/2023-01-18-high-variance…

Modified: January 23, 2023.

Tagged with:

managers are worst-case analyzers

There are a lot of difficult decisions to be made in life. Maybe you need to decide the business strategy of a company, knowing that good…

Modified: November 02, 2022.

Tagged with:

many models

An idea I got from [ John Higgs ]'s discussion of metamodernism is that taking [ all models are wrong ] to its logical conclusion requires…

Modified: January 06, 2023.

Tagged with: #modeling#ai#how-to-think

many selves

Sometimes I've been scared of losing my identity. In particular I worry about working a non-research job, or having sex with (or being…

Modified: February 07, 2022.

Tagged with:

marijuana

I have a private theory about what marijuana does. I'll try to articulate it here. I don't know much about the public theories, so maybe…

Modified: January 17, 2021.

Tagged with: #drugs

martingale

A martingale is any [ stochastic process ] that stays the same in expectation. Formally, is a martingale if This condition is related to…

Modified: August 27, 2022.

Tagged with: #math

massage

How to think about giving a good massage? Know which way the muscle fibers go. For deep release, exert force perpendicular to the muscle…

Modified: September 02, 2024.

Tagged with:

math

Modified: .

Tagged with: No Tags

matrix exponential

Reviewing this 3blue1brown video: https://www.youtube.com/watch?v=O85OWBJ2ayo The matrix exponential is written as E to the power of a…

Modified: November 20, 2023.

Tagged with: #math

matrix inversion lemma

The Woodbury-Morrison-Sherman matrix inversion lemma, is sometimes useful just for algebraic simplifications. In cases where and are…

Modified: March 16, 2022.

Tagged with: #math

matrix notation

Notation for Matrix Multiplication Let and . Then just by the definition of matrix multiplication (the summation over is performing…

Modified: March 16, 2022.

Tagged with:

maximal update parameterization

References: Hu, Yang (2022) Feature Learning in Infinite-Width Neural Networks https://arxiv.org/abs/2011.14522 Yang, Hu et al. (202…

Modified: December 29, 2023.

Tagged with:

maximum-entropy reinforcement learning

For any reward function and policy , consider the entropy-regularized reward Taking as our objective the (expected, discounted…

Modified: July 28, 2022.

Tagged with: #reinforcement-learning

mcmc notes

Note: these are personal notes, taken as I was refreshing myself on this material. They're mostly stream of consciousness and probably not…

Modified: March 16, 2022.

Tagged with: #math#machine-learning#bayes

measurable function

A function is measurable with respect to [ sigma-algebra ]s on its domain and on its range if the pre-image of any event is…

Modified: August 27, 2022.

Tagged with: #math

mechanistic interpretability

Modified: .

Tagged with: No Tags

meditation

The core insight that got me interested: "moments of recognizing your thoughts drifting and bringing them back to your breath" are not…

Modified: October 03, 2021.

Tagged with: #meditation#psychology

meditation following log

Apr 24, 2023 be clear on what you're doing in each session. don't mix concentration and emptiness I guess this is related to Tucker's…

Modified: December 13, 2023.

Tagged with: #meditation

meditation following log.sync-conflict-20240424-145224-QL2G4SK

Dec 13, 2023 Sangha session with Dustin: Something I realized during a concentration practice is that being honest about how ‘well’ the…

Modified: December 13, 2023.

Tagged with: #meditation

meditation ideas I resist

Generally I think the dharma is deeply true and that [ meditation ] done right is healthy and potentially very beneficial. But I struggle…

Modified: July 19, 2024.

Tagged with:

meditative attainments

Specific states or abilities that can arise from skillful meditation: feeling [ equanimity ] first cessation seeing nimitta accessing…

Modified: .

Tagged with: No Tags

melatonin

Modified: .

Tagged with: No Tags

memory

Modified: .

Tagged with: No Tags

memory-efficient attention

To train a [ transformer ] layer on a sequence of length requires the output of the attention computation where are matrices and is…

Modified: February 19, 2024.

Tagged with: #machine-learning

memory efficient backprop

Suppose we want to do [ automatic differentiation ] on a [ computational graph ] of sequential length . This could equally well be a…

Modified: January 02, 2024.

Tagged with:

memory reconsolidation

Described, among other places, in Unlocking the Emotional Brain . Insofar as much of Buddhism is about dissolving [ samskara ]s…

Modified: February 04, 2025.

Tagged with: #psychology#mental-health#neuroscience#meditation

mental models

One last thought mental models are so, so important. When I think about computer modeling. It's actually great computers are powerful they…

Modified: July 25, 2020.

Tagged with: #modeling#how-to-think

meritocracy

Like democracy , meritocracy is the worst form of social organization, except for all the others that have been tried. Of course it is good…

Modified: December 01, 2022.

Tagged with: #morality#how-to-think

mesa optimizer

References: Risks from Learned Optimization in Advanced Machine Learning Systems A [ reinforcement learning ] algorithm attempts to find the…

Modified: March 28, 2023.

Tagged with: #ai#reinforcement-learning

mescaline

Chemically, a substituted [ phenethylamine ]. Like [ dopamine ] but with methyl groups hanging from the two oxygens, and another oxygen…

Modified: December 21, 2022.

Tagged with: #drugs

meta learning

Generally this means training some aspect of the learning procedure itself. There is then an inner-loop learning procedure, which follows…

Modified: October 04, 2021.

Tagged with: #machine-learning

meta-level shape of machine learning

Unlike most modern [ deep learning ] systems, humans: don't have separate training/test phases (though we may have wake/[ sleep ]) don't…

Modified: January 16, 2022.

Tagged with: #machine-learning

meta-reasoning

Stuart Russell told the story of giving a talk on meta-reasoning at Stanford, with Don Knuth in the audience, where he opened with a slide…

Modified: October 17, 2022.

Tagged with: #ai

methamphetamine

n-methyl-[ amphetamine ]

Modified: July 24, 2023.

Tagged with:

metis

Modified: .

Tagged with: No Tags

metta

Pali (Buddhist) term for [ loving-kindness ]

Modified: April 10, 2024.

Tagged with:

middle way

From @visakanv on Twitter: (relevant to [ nothing matters ])

Modified: August 13, 2022.

Tagged with: #buddhism

mind at large

Modified: .

Tagged with: No Tags

mindfulness requires certainty

A lesson from [ Tucker Peck ]: unresolved questions are the worst thing in meditation. For example, you're just sitting down to practice…

Modified: November 27, 2023.

Tagged with: #meditation

minimax duality

Considering a bilevel optimization problem (or saddle point problem) on the two-argument function , in general it holds that That is, the…

Modified: July 07, 2022.

Tagged with: #math

minimum description length

Short descriptions of things, when they exist, must capture some kind of structure. The principle of [ Occam's razor ] posits that we should…

Modified: April 12, 2022.

Tagged with: #machine-learning#bayes

mirror descent

Mirror descent is a framework for optimization algorithms: many algorithms can be framed as mirror descent, and proofs about mirror descent…

Modified: October 03, 2020.

Tagged with: #machine-learning

mirror descent implementations

What pieces of [ mirror descent ] can we automate? See also [ natural gradient implementations ] Given a mirror function , we can compute…

Modified: September 07, 2020.

Tagged with:

mirror neurons

Modified: June 12, 2021.

Tagged with:

mission statement

(originally from 2020-04-29) On another note, last night I tried to dictate (on Otter) my sense of my life goals. I came up with a very…

Modified: January 24, 2022.

Tagged with: #personal

mitochondria

Modified: July 31, 2021.

Tagged with:

mixed effects

[ Otter notes ]: Can I explain what a mixed effects model is from a graphical model standpoint? On the inference side, I think it's just…

Modified: January 23, 2022.

Tagged with:

mixture of experts

A mixture-of-experts model consists of a set of functions , the 'experts', and a gating function that determines how to select which…

Modified: .

Tagged with: No Tags

mode-covering variational inference is incoherent

I have a [ strong opinion weakly held ] that doesn't seem to be wildly shared in the [ approximate Bayesian inference ] community: reverse…

Modified: March 14, 2022.

Tagged with: #machine-learning#bayes

model-agnostic meta learning

Original paper: Finn, Abbeel, and Levine, ICML 2017, https://arxiv.org/abs/1703.03400 An approach for [ meta learning ] that works with any…

Modified: February 20, 2022.

Tagged with:

model-based RL

Modified: .

Tagged with: No Tags

model-based rl

Often we don't explicitly use 'model-based RL' methods, instead people in robotics talk about Sim2Real: adapting a policy pretrained in a…

Modified: July 20, 2022.

Tagged with:

model integration

Modified: .

Tagged with: No Tags

molecular dynamics

Stack: goal: sample from conformations of arbitrary hydrocarbons (or whatever). simpler goal: sample from conformations of ethane. simpler…

Modified: May 16, 2020.

Tagged with: #physics#ideas

money supply

Naively you might think that the government just decides how many dollars there should be, and that's that. This is not true. Since [ IOUs…

Modified: February 09, 2022.

Tagged with: #finance

monoamine oxidase

A monoamine oxidase (MAO) is an enzyme that breaks down mono-[ amine ] neurotransmitters such as [ dopamine ], [ serotonin…

Modified: May 22, 2022.

Tagged with: #drugs#organic-chemistry

monte carlo tree search

A very natural form of [ meta-reasoning ] that selects the most promising computations. The simplest form of 'expanding' a node assumes a…

Modified: March 22, 2022.

Tagged with: #ai

moral realism

There is a connection between moral realism and belief in [ qualia ]. If you see "experience" ([ awareness ]) as a real, fundamental aspect…

Modified: .

Tagged with: No Tags

morning person

I don't hold the moral view that it's better to be a morning person than an evening person. Having always tended towards a later sleep…

Modified: March 14, 2023.

Tagged with: #personal#life-advice

most learning is by demonstration

In any human-to-human interaction, language carries some very important high-order bits, but it can only carry a few bits. It can help…

Modified: June 12, 2021.

Tagged with: #teaching#machine-learning

most people don't care

This is one of the big problems with the world. Not the only one, and not the only way to look at it. But it's everywhere. status : a…

Modified: January 25, 2022.

Tagged with: #how-to-think#mental-health

most stupid ideas are stupid ideas

Modified: .

Tagged with: No Tags

most work is bullshit

(see David Graeber https://www.strike.coop/bullshit-jobs/ ) Most work is oriented towards achieving [ instrumental goal ]s. But most…

Modified: February 25, 2022.

Tagged with: #how-to-think

motivation

I like this take on working with procrastination from a [ nondual ] [ awareness ] perspective: From the viewpoint of "the beyond…

Modified: June 21, 2024.

Tagged with:

motorbike tips

maneuvering: the bike goes where I look. look around the turn I want to do. keep elbows up. shift body weight to counterbalance the bike. E…

Modified: May 12, 2022.

Tagged with:

multimodal transformer

possible refs: google's multimodal architectures: https://webcache.googleusercontent.com/search?q=cache:https://towardsdatascience.com…

Modified: September 25, 2023.

Tagged with:

multiplicative interaction

From a conversation I had about [ attention ] mechanisms in deep architectures. Maybe that terminology is too suggestive --- it's just a…

Modified: March 03, 2024.

Tagged with: #ai#machine-learning

multivariate gaussian

We say that a random vector is multivariate Gaussian with mean and covariance matrix if it can be written where is a vector if i.i.d…

Modified: March 16, 2022.

Tagged with: #math

multivariate time series

[ thoughts on multivariate causalimpact ]

Modified: February 15, 2022.

Tagged with: #modeling#bayes

mutually orthogonal communities

This was originally a section of breakup.org, written several years ago. this is more related to jobs and identity, but for cases when I get…

Modified: May 22, 2021.

Tagged with:

my goals

I want to intentionally spend my time well. I remember back in grad school I would spend evenings reading papers, just as a form of growth…

Modified: February 25, 2022.

Tagged with: #personal

my relationship with tech

I've identified as a 'tech' person, but I now feel uncomfortable in many tech circles. What is tech and what does it mean to be a tech…

Modified: January 24, 2022.

Tagged with: #personal

my values

It's a useful exercise to occasionally reflect on what I value. stab 1: Generally pro tech, creating new things, non-zero-sum contributions…

Modified: November 27, 2023.

Tagged with: #personal

myelin

Modified: .

Tagged with: No Tags

nasty, brutish, and short

Modified: February 09, 2020.

Tagged with:

nattokinase

Recommended by Michael Edward Johnson:

Modified: September 28, 2023.

Tagged with:

natural abstraction

A 'natural' abstraction is one that we expect any agent (or at least, a wide range of agents) to develop because it gets at something…

Modified: May 04, 2023.

Tagged with:

natural experiment

Modified: .

Tagged with: No Tags

natural gradient

We don't typically think of it this way, but you can derive a [ gradient descent ] step as finding the point that minimizes a linearized…

Modified: July 06, 2022.

Tagged with: #machine-learning

natural gradient implementations

How can we automate [ natural gradient ]? See also [ mirror descent implementations ]

Modified: September 19, 2020.

Tagged with:

nearest neighbor

Cool trick: some applications can improve on nearest-neighbor lookup by training 'Exemplar SVM's. Instead of matching against a set of…

Modified: April 15, 2023.

Tagged with:

negative utilitarianism

Modified: .

Tagged with: No Tags

negative utility

My position (a [ strong opinion weakly held ]) is that global utility is currently negative, and probably always has been. It's conceivable…

Modified: August 25, 2022.

Tagged with: #morality

negligible

A negligible function is a function such that, for any positive integer there exists an integer such that for all , i.e., that…

Modified: October 23, 2022.

Tagged with: #math#crypto

nested SMC

Christian Naesseth, Fredrik Lindsten, Thomas Schon (2015): http://proceedings.mlr.press/v37/naesseth15.html The main idea: In an SMC…

Modified: July 14, 2021.

Tagged with: #machine-learning#papers

neural nets do work

Like the proverbial half-full glass, smart people can look at the same reality of the current capacities of neural nets, and come to…

Modified: April 07, 2020.

Tagged with: #machine-learning#personal

neural nets don't just interpolate

Sometimes you'll see people say that neural nets 'just' memorize and interpolate their training data. No one denies that neural nets with…

Modified: .

Tagged with: No Tags

neuron

Parts of a neuron: dendrites: these branch out to receive connections from other cells axons: these branch out to send signals to other…

Modified: August 08, 2021.

Tagged with: #biology#neuroscience

neurotransmitter

Modified: .

Tagged with: No Tags

nihilism

Modified: February 25, 2022.

Tagged with:

no free lunch theorem

The folklore no-free-lunch 'theorem' in machine learning says that, for any pair of learning algorithms, there exists some dataset on which…

Modified: March 04, 2022.

Tagged with: #machine-learning

no plan survives contact with the enemy

Modified: .

Tagged with: No Tags

no-self

No-self is one of the [ three characteristics ] that traditional Buddhism holds are present in all phenomena. In later Buddhism, the…

Modified: May 30, 2023.

Tagged with: #buddhism#meditation#drugs

noisy natural gradient as VI

https://arxiv.org/abs/1712.02390 Basic idea: optimizers like Adam and RMSProp already keep track of posterior curvature estimates. These are…

Modified: October 30, 2020.

Tagged with: #machine-learning#papers

nominal GDP target

Instead of directly targeting a specific rate of inflation, a [ central bank ] may target a fixed rate of nominal GDP growth, which is equal…

Modified: .

Tagged with: No Tags

non-dominating force

One way to model real-world [ causality ] is a bunch of forces working with and against each other. In this view, no individual force…

Modified: July 14, 2023.

Tagged with: #how-to-think

non-fungible token

NFTs 101: https://medium.com/@intenex/nfts-101-why-nfts-are-a-generational-innovation-4626ae803e3b Among many other things, NFTs are…

Modified: October 05, 2021.

Tagged with: #finance

non-player character

Modified: .

Tagged with: No Tags

nondual

Modified: .

Tagged with: No Tags

nootropics

Obligatory disclaimer: there will never be a drug to turn you into Einstein. Most of effective high-level thinking lies in 'software…

Modified: July 09, 2022.

Tagged with: #neuroscience

norepinephrine

Modified: August 08, 2021.

Tagged with: #neuroscience

normalized advantage function

References: Gu et al., Continuous Deep Q-Learning with Model-based Acceleration (2016). Instead of modeling directly, we build a network…

Modified: July 19, 2022.

Tagged with: #reinforcement-learning

not true enough

Something can be true but not 'true enough'. That is, you have a compelling causal theory for why X should increase Y. It might be that the…

Modified: August 21, 2020.

Tagged with: #how-to-think

notes on Hamming

I've started reading The Art of Doing Science and Engineering by Richard Hamming. History of computing: Analog computing goes back forever…

Modified: June 03, 2020.

Tagged with:

nothing matters

Because: [ goals are arbitrary ]: achieving a goal, or failing to, doesn't really matter because the goal was arbitrary anyway. From the…

Modified: February 25, 2022.

Tagged with: #morality

nothing to do

There's a spiritual idea, in Buddhism and elsewhere, that there is "nothing to do": everything is already suffused with "primordial…

Modified: June 06, 2023.

Tagged with: #buddhism#meditation

nucleophile

Modified: August 02, 2020.

Tagged with: #chemistry

nucleotide

Modified: February 10, 2022.

Tagged with: #biology#chemistry

nucleus sampling

Modified: .

Tagged with: No Tags

numerics

Don't invert that matrix: https://www.johndcook.com/blog/2010/01/19/dont-invert-that-matrix/ Seven sins of numerical linear algebra…

Modified: December 28, 2022.

Tagged with: #computer-science

objectives are big

A very incomplete and maybe nonsensical intuition I want to explore. Classically, people talk about very simple [ reward ] functions like…

Modified: March 31, 2023.

Tagged with: #ai#reinforcement-learning#alignment

off-policy

A few (relatively uninformed) thoughts about on- vs off-policy [ reinforcement learning ]. Advantages of on-policy learning: On-policy…

Modified: April 23, 2022.

Tagged with: #reinforcement-learning

old daily templates

Original: Daily reflections What am I grateful for today?:: Some goals : Goals for the next ~year:: Goals for the next ~month:: Goals for…

Modified: January 23, 2022.

Tagged with:

on-policy learning

Modified: .

Tagged with: No Tags

one taste

The brain doesn't have separate models of each of the [ sense gate ]s (and thought). Instead it just stores each moment of perception as a…

Modified: .

Tagged with: No Tags

one-way function

Informally, a function is a one-way function if it is easy to compute but hard to invert. Or more generally, hard to pseudo-invert, i.e…

Modified: October 23, 2022.

Tagged with: #crypto

ongoing projects

These are things that I might plausibly decide I want to work on when I sit down on the weekend. Expanding nodes on this graph. Blogging…

Modified: February 22, 2020.

Tagged with: #personal#ideas

ontological crisis

How do we maintain values when our models of the world shift? If someone's goal in life is to "do God's will", and then they come to believe…

Modified: April 12, 2023.

Tagged with: #alignment#ai

optimism

As Josh Marshall said , at the beginning of the Trump presidency: "Optimism is not primarily a prediction but an ethic, a philosophy, a way…

Modified: June 08, 2021.

Tagged with: #how-to-think

option

Modified: October 26, 2021.

Tagged with: #finance

optional stopping

If is a [ martingale ] and is a [ stopping time ], then any of the following conditions implies that : The stopping time is bounded…

Modified: August 29, 2022.

Tagged with: #math

organic chemistry

Modified: May 01, 2020.

Tagged with:

origin of suffering

Ken McLeod claims that 'emotional reactivity' is the origin of suffering. Pain consists both in what happens and in our reaction to it. But…

Modified: October 06, 2021.

Tagged with: #buddhism

overparameterize

Modified: March 02, 2022.

Tagged with:

ownership

Modified: .

Tagged with: No Tags

oxidation

mnemonic: OIL RIG = 'oxidation is losing (electrons), reduction is gaining (electrons)' in contrast to [ acid-base chemistry ], which is…

Modified: July 31, 2021.

Tagged with: #chemistry

oxidative phosphorylation

This is how [ mitochondria ] produce most of their [ ATP ]. Mitochondria have an outer membrane and an inner membrane, so there are two…

Modified: July 31, 2021.

Tagged with: #biology

p-zombie

Modified: .

Tagged with: No Tags

pale blue dot

Look again at that dot. That's here. That's home. That's us. On it everyone you love, everyone you know, everyone you ever heard of, every…

Modified: November 30, 2022.

Tagged with:

paperclip maximizer

Modified: .

Tagged with: No Tags

papers to read

Modified: July 10, 2020.

Tagged with: #papers

partial differential equation

References for PDEs: commutant's Youtube videos: https://www.youtube.com/playlist?list=PLF6061160B55B0203 Fundamental PDEs wave equation…

Modified: June 07, 2024.

Tagged with:

particle MCMC

Basic notes from https://www.stats.ox.ac.uk/~doucet/andrieu_doucet_holenstein_PMCMC.pdf Setup: we have parameters and time series model…

Modified: April 06, 2020.

Tagged with:

party ideas

Chocolate tasting: buy a bunch of high-end, single-origin chocolate bars. Parcel them out blind. Give people a pad to take notes on what…

Modified: May 19, 2020.

Tagged with: #ideas

penalties are constraints

We often see optimization problems with objectives of the form where is the main function of interest (e.g., training loss in machine…

Modified: July 15, 2022.

Tagged with:

people like hearing their name

“Remember that a person’s name is to that person the sweetest and most important sound in any language.” Dale Carnegie (How to Win Friends…

Modified: March 02, 2022.

Tagged with: #relationships

people want to see you thrive

When you're thinking about doing something that feels right to you, it's easy to get caught up in worrying about what other people will…

Modified: February 10, 2022.

Tagged with: #life-advice

perceiver

reading the perceiver papers from Deepmind: Perceiver: Jaegle et al 2021 https://arxiv.org/abs/2103.03206 Perceiver-IO: Jaegle et al 202…

Modified: September 25, 2023.

Tagged with: #ai#machine-learning

persistent hallucination

In the [ 5-MeO-DMT ] trip where I experienced [ ego death ], I saw a [ magical display ] of beautiful colors and flowing motion and…

Modified: February 10, 2022.

Tagged with: #drugs

personal AI Effect

The AI Effect refers to the widely-recognized phenomenon that 'once we know how to do it, it's not AI'. For example, playing chess well…

Modified: May 29, 2020.

Tagged with: #ai

personal philosophy

I always found it weird that philosophy spends so much time talking about specific historical philosophers. Who cares what Aristotle, or…

Modified: January 24, 2022.

Tagged with: #teaching#philosophy

personal value-over-replacement

When considering one's impact on the world, it's important (? or at least tempting) to think about about your value-over-replacement. If you…

Modified: July 07, 2023.

Tagged with:

phase change hypothesis

(see also: [ large models ]) There's a viewpoint that neural nets just memorize the training data, so the more training data you have, the…

Modified: February 10, 2022.

Tagged with: #machine-learning#modeling

phase space

Modified: .

Tagged with: No Tags

phase transition

Modified: .

Tagged with: No Tags

phenethylamine

Modified: May 14, 2021.

Tagged with: #chemistry#drugs

phenibut

Developed and widely used in Russia, phenibut is an analogue of [ GABA ] with a phenyl ring substituted at the carbon, giving it the name…

Modified: September 28, 2023.

Tagged with:

phosphate

Why Nature Chose Phosphates (science.org)

Modified: January 19, 2022.

Tagged with: #biology#chemistry

poems

To His Coy Mistress Andrew Marvell, 1681 Had we but world enough and time, This coyness, lady, were no crime. We would sit down, and think…

Modified: July 19, 2024.

Tagged with:

pointing out

The paradoxical thing about pointing-out style meditation teaching is that you can't really explain the instructions when they're unclear…

Modified: October 30, 2021.

Tagged with: #meditation

polar

Modified: .

Tagged with: No Tags

policy

Modified: March 02, 2022.

Tagged with:

policy gradient

(see also my [ deep RL notes ] from John Schulman's class several years ago, which cover much of the same material) We can approach…

Modified: March 14, 2024.

Tagged with: #reinforcement-learning

polyak averaging

Modified: .

Tagged with: No Tags

positional embedding

There are a few ways to do this. Google's PaLM uses rotary embeddings so it seems like that's probably close to the state of the art? But…

Modified: September 28, 2023.

Tagged with: #transformers#ai

positive sum

Modified: .

Tagged with: No Tags

potential outcomes

Different experimental conditions may give rise to different outcomes . For example, let the variable indicate whether a person is…

Modified: August 06, 2021.

Tagged with: #causality

prayer is therapy

Prayer is a form of [ therapy ]. It's about clarifying your values: figuring out what you really want so that you can ask God for it. and…

Modified: February 07, 2022.

Tagged with:

predictable process

A [ stochastic process ] is predictable if its value at time is fully determined by information available at time . Any fully…

Modified: August 27, 2022.

Tagged with: #math

prediction as a model-building exercise

A really valuable exercise that I should consider building into my routine is to regularly try to make and write down explicit predictions…

Modified: January 24, 2021.

Tagged with: #how-to-think

predictive agent

Consider an agent that is purely concerned with [ predictive processing ]: finding the optimal [ compression ], or equivalently the optimal…

Modified: April 12, 2023.

Tagged with: #ai

predictive processing

The theory of predictive processing seems to be attracting a lot of interest in neuroscience and [ meditation ] circles. I want to try to…

Modified: .

Tagged with: No Tags

preference cascade

https://www.quora.com/What-is-a-preference-cascade A lot of how people act is driven by how they think they're 'supposed' to act. There's…

Modified: February 22, 2022.

Tagged with:

previously read

AI / RL Distributional RL book: https://www.distributional-rl.org/ Alignment Sequences: Value learning: https://www.alignmentforum.org/s…

Modified: June 27, 2023.

Tagged with: #papers#ideas

principal-agent problem

Modified: .

Tagged with: No Tags

priors are conceptual attention

A Bayesian view of (one aspect of) [ attention ] inspired by a conversation with Shamil Chandaria on [ predictive processing ]. (but this…

Modified: May 25, 2023.

Tagged with: #ai

privacy

It seems like there is, or can be, a virtuous relationship between privacy and generalization. You don't want to memorize too many…

Modified: February 14, 2021.

Tagged with: #machine-learning

privilege

Illegible privilege We often talk about the 'privilege' associated with certain categories: being born white, straight, male, rich, in a…

Modified: April 04, 2024.

Tagged with: #how-to-think#lgbt

pro-social identity

(I got this concept from SuccessfulFriend.) As people grow up and form their identities, they need models, and not just models; they need…

Modified: February 22, 2022.

Tagged with: #growing-up#psychology

probabilistic program induction

Can we think about [ generative flow network ]s as a potentially tractable formulation of probabilistic program induction?! executing a line…

Modified: March 14, 2022.

Tagged with: #machine-learning#ai

probabilistic programming

Modified: February 08, 2020.

Tagged with: #machine-learning#modeling

probabilistic programming is not AI research

Many [ probabilistic programming ] researchers frame their work as part of the broader problem of [ artificial intelligence ]. Artificial…

Modified: December 01, 2023.

Tagged with: #machine-learning#bayes

probabilistic transformers

A short note on interpreting a transformer layer as performing maximum-likelihood inference in a Gaussian mixture model: https://arxiv.org…

Modified: October 30, 2020.

Tagged with: #papers#machine-learning

probabilities hide detail

Matt Levine explains how a financier might react to losing a billion dollars: Sure sure the risks didn’t work out but you probably have a…

Modified: February 22, 2022.

Tagged with: #modeling

probability space

A probability space consists of: A set of outcomes aka possible worlds; these represent all the ways the world might be. This is the…

Modified: August 27, 2022.

Tagged with: #math

process is frequentist

(aka, why frequentists will always make more money) In the "real" (corporate/governmental) world, most high-level decision making is…

Modified: March 04, 2022.

Tagged with: #bayes

procrastination

Modified: September 27, 2024.

Tagged with: #effectiveness#mental-health