All Notes: Nonlinear Function

All Notes

anion

Modified: .

Tagged with: No Tags

anonymized names

In the published version of these notes, names are anonymized to protect the privacy of friends and colleagues.

Modified: February 07, 2022.

Tagged with:

approximate Bayesian inference

Modified: December 01, 2023.

Tagged with: #machine-learning#math

argument

Modified: January 23, 2022.

Tagged with:

arguments can't be trusted

Most [ argument ] that people engage in isn't formally persuasive. It's a back and forth of giving reasons: I give a reason to do a thing…

Modified: January 23, 2022.

Tagged with: #how-to-think#growing-up

articulation regularization

see also [ communication is processing ] Developing an idea with someone else applies regularization to the idea. It guarantees that the…

Modified: March 04, 2022.

Tagged with: #relationships#research#how-to-think

artificial intelligence

Intelligence is what makes humans special. PhdAdvisor defines it as the ability to make high-quality decisions. I think we can distinguish…

Modified: March 02, 2022.

Tagged with: #ai

asian glow

References: https://en.wikipedia.org/wiki/Alcohol_flush_reaction There seem to be three genes associated with poor alcohol tolerance…

Modified: January 21, 2023.

Tagged with: #chemistry

asian movies and TV

To watch: The Untamed: https://www.youtube.com/watch?v=SlpEVsKiYCU HIStory (Taiwanese boy love) HisTory 2: right or wrong: https://www…

Modified: May 16, 2021.

Tagged with:

ask for evidence

In a work context: Consider a junior engineer and a senior engineer. The junior engineer says, 'I think doing X will be better than doing Y…

Modified: February 10, 2022.

Tagged with: #how-to-think#growing-up

associative memory

[ Hopfield network ] [ sparse distributed memory ] Milledge et al. (2022), Universal Hopfield Networks: A General Framework for Single-Shot…

Modified: March 29, 2024.

Tagged with:

attachment

Modified: February 25, 2022.

Tagged with:

attention

One of the best ideas in machine learning. (I even thought so in 2011!) There are two common mechanisms: 'soft' and 'hard'. In both cases…

Modified: January 24, 2022.

Tagged with: #machine-learning#ai#meditation#psychology#neuroscience

attention and utility

Thesis: [ attention ] is the dominant factor in our utility. It follows that work that shifts our attention can genuinely improve global…

Modified: February 25, 2022.

Tagged with:

authentic

A paradox maybe worth remembering: if I want to build a conscious identity around something, then I have to be willing to feel inauthentic…

Modified: September 27, 2021.

Tagged with:

automatic differentiation

This is my stab at explaining automatic differentiation, specifically backprop and applications to neural nets. A few dimensions to think…

Modified: August 23, 2022.

Tagged with: #math#machine-learning

autoregressive

Modified: .

Tagged with: No Tags

aversion

The most important thing to know about aversion (avoidance / fear / ...) as an emotional response is that it's, at best, exponentially less…

Modified: April 20, 2024.

Tagged with: #fundamental#meditation#life-advice

awakened awareness

[ Dan Brown ] uses this term a lot. Loosely, I think it means [ awareness ] that is aware of itself, without [ fabrication ]. Like…

Modified: March 23, 2023.

Tagged with: #meditation

awakening

Here's a perspective I like on spiritual awakening, elaborating on a metaphor used by [ Michael Taft ]. Suppose you awaken from a dream: a…

Modified: March 23, 2023.

Tagged with: #meditation#buddhism

awareness

As investigated in [ nondual ] or [ mahamudra ] meditation practice, awareness is both the 'field' in which all of experience occurs, and…

Modified: March 23, 2023.

Tagged with: #meditation

basket of options

Matt Levine points out a basic fact of derivatives theory: a basket of options is worth more than an option on a basket. If five stocks…

Modified: December 04, 2023.

Tagged with: #finance

be open to friendship

I've sometimes felt a scarcity mindset with respect to friendship. Like: friends and especially close friends and relationships are things…

Modified: February 10, 2022.

Tagged with: #personal

beam search

Modified: .

Tagged with: No Tags

before we meet

What would I say to someone I met up with on Tindr? How do I think about the encounter? I like almost everyone, so in particular I like you…

Modified: November 28, 2021.

Tagged with: #personal

behavioral cloning

Modified: .

Tagged with: No Tags

being an undergrad

just in case I'm ever a professor/teacher and forget the personal side of what people might be going through: as a freshman: I started out…

Modified: February 10, 2022.

Tagged with: #teaching#growing-up

being good is a privilege

Imagine a really good person: someone whom everyone likes, is warm and friendly towards everyone they meet, gives freely of themselves, has…

Modified: April 01, 2022.

Tagged with: #morality

being yourself takes practice

Related to: [ generative vs discriminative modeling ] The difficulty of [ learning new skills ] I remember learning to play the violin when…

Modified: May 01, 2023.

Tagged with: #life-advice

believe in something

I used to believe in AI and machine learning. It was obvious to me as a student that the ability to create intelligent machines will be a…

Modified: September 18, 2020.

Tagged with:

benzodiazepines

Modified: .

Tagged with: No Tags

berkeley eecs graduate admissions

This year (2014-15) I served as a student reader on the Berkeley EECS PhD admissions committee. Berkeley typically gets about N…

Modified: March 06, 2022.

Tagged with: #academia

best explanations

lagrange multipliers: www.eecs.berkeley.edu/~klein/papers/lagrange-multipliers.pdf legendre transfors / convex duality: http://student…

Modified: February 10, 2022.

Tagged with: #teaching

best of all possible worlds

Modified: .

Tagged with: No Tags

bias-variance tradeoff

I think of "variance" as the error in a statistical estimate that comes from not having enough data (assuming an [ identifiable ] model…

Modified: September 28, 2023.

Tagged with: #machine-learning

bids

The Atlantic has this article on "Masters of Love" http://www.theatlantic.com/health/archive/2014/06/happily-ever-after/372573/ presenting…

Modified: February 10, 2022.

Tagged with: #relationships

biggest problems in AI

I think this might be the wrong way to frame this. The field is CS, and the biggest problem in CS is AI. Within AI, the problems are…

Modified: September 07, 2020.

Tagged with:

bitter lesson

http://www.incompleteideas.net/IncIdeas/BitterLesson.html The bitter lesson is based on the historical observations that 1) AI researchers…

Modified: July 25, 2023.

Tagged with: #ai

blackpill dynamics

(followup q: is 'blackpill' the best name? or is that a broader concept, there are blackpills other than the thing I'm describing, so maybe…

Modified: .

Tagged with: No Tags

blockchain

Modified: .

Tagged with: No Tags

blog posts to write

[ writing inbox ] the privilege of having correct advice work out for you ([ the privilege of advice working out ]) Let's stipulate that any…

Modified: February 10, 2022.

Tagged with: #personal#ideas

blur your eyes

Jitendra Malik would often tell us that some idea or explanation makes sense if you 'blur your eyes'. This seems counterintuitive, but it's…

Modified: January 24, 2022.

Tagged with:

body practice

Possible resources on tai chi, qi gong Paul Lam beginner video: https://www.youtube.com/watch?v=rbFzbS2lLT0 if it's good, can pay for his…

Modified: .

Tagged with: No Tags

book goals

[ meditation ] and dharma: The Buddha's Own Words Mastering the Core Teachings of the Buddha Science of Enlightenment Seeing that Frees…

Modified: May 20, 2022.

Tagged with: #personal

books I recommend

Fiction memoirs of hadrian shogun red plenty name of the wind Cloud Atlas Lord of Light Tomorrow and tomorrow and tomorrow gay: Giovanni's…

Modified: July 05, 2022.

Tagged with: #life-advice

bounded cognition

Modified: .

Tagged with: No Tags

brahmavihara

The four qualities of mind that it is impossible to have too much of, aka the four 'divine abodes' or immeasurables: [ loving-kindness…

Modified: May 19, 2022.

Tagged with: #buddhism

brain

Sources: wikipedia GPT-4 https://ocw.mit.edu/courses/9-13-the-human-brain-spring-2019/video_galleries/lecture-videos/ Major parts of the…

Modified: October 27, 2023.

Tagged with:

bupropion

Antidepressent that acts as a [ norepinephrine ]-[ dopamine ] reuptake inhibitor. In the class of substituted [ phenethylamine ]s and (more…

Modified: August 27, 2022.

Tagged with: #mental-health#neuroscience

calibration

A nice paper that gets at some subtleties of calibration: Daniel D. Johnson, Daniel Tarlow, David Duvenaud, Chris J. Maddison. Experts Don't…

Modified: March 26, 2024.

Tagged with: #modeling#machine-learning

cannabinoid

Endocannabinoids are retrograde [ neurotransmitter ]s, meaning that they pass 'backwards' (from dendrite to axon) through the synaptic cleft…

Modified: August 24, 2022.

Tagged with: #neuroscience

capabilities research

In the discourse around [ AI safety ] you sometimes see the claim that research on AI capabilities is harmful to the extent that it outpaces…

Modified: February 26, 2022.

Tagged with: #ai

carbonyl

Modified: February 10, 2022.

Tagged with: #chemistry

carbonyl group

Modified: August 02, 2020.

Tagged with:

care about my work enough to want to get better

The problem with work that's 'just a job' is that you'll never be as good at it as at work that really excites you, where you really want…

Modified: January 25, 2022.

Tagged with: #effectiveness

cargo-culting

Modified: .

Tagged with: No Tags

carve the world at the joints

Modified: January 23, 2022.

Tagged with:

casual love

aka 'little love' or 'big love' or various other things Saved from https://www.carsieblanton.com/blog/post/82149148832/casual-love This…

Modified: February 22, 2022.

Tagged with: #life-advice

catastrophic forgetting

Modified: .

Tagged with: No Tags

cation

Modified: .

Tagged with: No Tags

causal graph

Modified: .

Tagged with: No Tags

causal inference

How do we infer [ causality ] from observational data? This question is important in science and is closely related to the progress of [ off…

Modified: July 05, 2022.

Tagged with: #causality

causality

see [ causal inference ] Great Causality & ML Papers and Researchers – Blog (logangraham.xyz)

Modified: August 02, 2021.

Tagged with: #causality

cellular respiration

We can view cellular respiration (and combustion more generally) from a high level as the transfer of electrons from carbon to oxygen atoms…

Modified: July 31, 2021.

Tagged with:

central bank

Modified: March 13, 2022.

Tagged with: #finance

cerebellum

Modified: .

Tagged with: No Tags

cessation of suffering

see: https://www.abolitionist.com/ https://qualiacomputing.com/2018/11/07/anti-tolerance-drugs/ [ suffering ] may seem inevitable. [ karma…

Modified: June 14, 2023.

Tagged with: #buddhism

ceteris paribus

Other things held equal. Abstract reasoning is good for coming to 'ceteris paribus' conclusions. It's easy to identify one force acting in…

Modified: May 23, 2021.

Tagged with:

chain of thought

Modified: .

Tagged with: No Tags

chain rule

There are two major 'chain rules' relevant to machine learning: the chain rule of probability theory and the chain rule from calculus…

Modified: August 13, 2022.

Tagged with: #math

change-signaling event

Say you want to make a big personal change: to stop smoking, or to stop eating meat, or to meditate every day, or introduce yourself to a…

Modified: June 07, 2021.

Tagged with: #psychology

choicefulness

I got this concept from Matt Johnson. People overestimate what they can do in a week, but underestimate what they can do in ten years. To do…

Modified: January 24, 2022.

Tagged with: #how-to-think

circuit complexity

Basic classes: NC (Nick's class) : Boolean circuits with fan-in 2 AC (alternating circuits) : Boolean circuits with unlimited fan-in for AND…

Modified: .

Tagged with: No Tags

citric acid cycle

Also known as the Krebs cycle. Is the final common pathway for oxidation ('burning') of fuel molecules: carbs, fats, proteins. The fuel…

Modified: July 31, 2021.

Tagged with: #biology

classic papers

More Is Different (kit.edu) On proof and progress in mathematics (Thurston, 1994) On Being the Right Size (Haldane, 1928))

Modified: March 02, 2022.

Tagged with: #papers#fundamental

classification is special

The [ distinction ] between classification and regression is, from one point of view, arbitrary: it's all just function approximation, and…

Modified: March 12, 2021.

Tagged with: #machine-learning

coarse-to-fine

Dan Klein once said I don't remember the context, but I took his NLP class in spring 2011 so it was probably around then. that he thought…

Modified: .

Tagged with: No Tags

cognitive structure

[ conceptual scaffolding ]

Modified: April 29, 2020.

Tagged with:

cognitive technology

Modified: January 07, 2022.

Tagged with: #how-to-think

come alive

"find the thing that makes you come alive, and do that." I find myself thinking about 'giving up' in the sense of dreams of being an…

Modified: February 10, 2022.

Tagged with: #life-advice#growing-up

coming out

Modified: February 07, 2022.

Tagged with:

commitment scheme

A commitment scheme allows for one party to publicly commit to some value without revealing that value. For example, Alice wants to bet on…

Modified: October 23, 2022.

Tagged with: #crypto

common knowledge

Modified: January 09, 2021.

Tagged with: #how-to-think

communication is processing

Talking and writing are not nearly as much about communication as we think. That's part of it, of course. But a significant portion, often…

Modified: March 04, 2022.

Tagged with: #how-to-think

complacent

Modified: February 07, 2022.

Tagged with:

complete the square

Multivariate Completion of Squares A useful trick: if is a symmetric, nonsingular matrix, then This is easy to see just by expanding out…

Modified: March 16, 2022.

Tagged with: #math

compositional natural gradient

TO READ: Kronecker factored Approximate Curvature : K-FAC Practical Gauss-Newton : gives recursions for computing Hessian blocks, and…

Modified: October 30, 2020.

Tagged with:

compression

(below was originally an email to SuccessfulFriend, copying here for posterity) learning as compression seems like kind of a folk idea with…

Modified: September 06, 2021.

Tagged with: #machine-learning#how-to-think

computation is important

Arguably the core insight of deep learning / [ differentiable program ]ming is that the shape and structure of the computations we do are so…

Modified: September 13, 2022.

Tagged with: #machine-learning#ai

computational complexity

Modified: .

Tagged with: No Tags

computational functionalism

The view that performing the "right kind of computation" is necessary and sufficient for [ consciousness ]. Chalmers (1995): if a person's…

Modified: September 21, 2023.

Tagged with:

computational graph

Modified: .

Tagged with: No Tags

computational lens notes

**from “theory of computation as a lens on the sciences” ehud kalai talks about structural robustness in games: an equilibrium is robust if…

Modified: February 07, 2022.

Tagged with:

computational life coach

How do you start building and selling [ computational therapy ]? It can't just be a medical product, because that's a hugely regulated and…

Modified: February 10, 2022.

Tagged with: #ideas#ai

computational therapy

See also: [ computational life coach ] A recurring dream I have is to use AI to solve mental health. It is simultaneously one of the most…

Modified: May 16, 2022.

Tagged with: #ideas#ai

concentration

Concentration is something of a misnomer for the meditative practice and states of samādhi . The word "concentration" carries connotations…

Modified: March 23, 2023.

Tagged with: #meditation

concentration inequalities

Modified: .

Tagged with: No Tags

conceptual chunks

Modified: July 08, 2020.

Tagged with:

conceptual scaffolding

Related: Learning increasingly complex ideas may amount to forming larger effective chunk sizes Expertise requires increasingly…

Modified: May 04, 2020.

Tagged with: #how-to-think

confidence all the way up

Nate Soares wrote an essay a while ago on how he experiences [ self-confidence ] when opining on difficult topics: On reflection, I've…

Modified: July 25, 2023.

Tagged with: #life-advice

confidence is about accepting failure

https://markmanson.net/how-to-be-confident see also [ imposter syndrome ] and the thesis that a lack of confidence is often an aversion to…

Modified: July 25, 2023.

Tagged with: #life-advice

connection

Modified: .

Tagged with: No Tags

connections between DDPG and Bayesian optimization

Modified: .

Tagged with: No Tags

consciousness

Philosophical views on consciousness: Buddhist and meditative traditions focus on [ awareness ]. They claim that consciousness has nothing…

Modified: July 14, 2023.

Tagged with: #meditation#ai

conspiracy is a thing for a reason

Multiple people working together are fundamentally more powerful than a single person working alone. Governments recognize this fact, which…

Modified: March 04, 2022.

Tagged with: #life-advice#how-to-think

constrained optimization

Suppose we want to optimize an objective under some equality and/or inequality constraints, Some general classes of approach we can use are…

Modified: July 07, 2022.

Tagged with: #math#machine-learning

constraints can be good

All else equal, constraints prevent you from doing what you'd have otherwise wanted to do, which is bad. But. Constraints prevent [ analysis…

Modified: January 09, 2021.

Tagged with: #how-to-think

continuous chain of thought

it seems clear that forcing [ transformer ]s to produce discrete tokens is a significant constraint on their reasoning ability. tokens…

Modified: .

Tagged with: No Tags

continuous structure learning

Relevant papers: DIfferentiable compositional kernel learning for Gaussian Processes (Sun et al., 2018) Differentiable Architecture Search…

Modified: March 07, 2020.

Tagged with: #papers#machine-learning#bayes

contraction

A contraction mapping on a metric space is a function such that for all and for some , called the [ Lipschitz ] constant of the map…

Modified: August 13, 2022.

Tagged with: #math

contrastive divergence

A method for fitting an unnormalized probability density (aka [ energy-based model ]) to data. Note that this is a different and harder…

Modified: May 15, 2021.

Tagged with: #machine-learning

contrastive learning

A technique for [ representation ] learning in which semantically similar datapoints are encouraged to have similar representations, and…

Modified: April 30, 2022.

Tagged with: #machine-learning

control variate

Modified: January 16, 2022.

Tagged with: #machine-learning

conversation as a game

Okay so there’s a lot of research on what conversations are, what the goals are (of course I don’t know most of this research…). It seems as…

Modified: February 13, 2022.

Tagged with: #ai#how-to-think

convex

A convex function satisfies the property that a line between any two points on its graph is on or above the graph: for any . It is…

Modified: July 09, 2022.

Tagged with: #math

convex dual

See also: https://www2.sonycsl.co.jp/person/nielsen/Note-LegendreTransformation.pdf Jess Riedel on the Legendre transform in physics looks…

Modified: June 25, 2022.

Tagged with: #math

convolution

Modified: .

Tagged with: No Tags

cooking technique

Order of operations When should you add oil when preparing to sauté? Generally the answer is "after preheating the pan, unless it's nonstick…

Modified: March 30, 2022.

Tagged with:

cooperative game

References: https://en.wikipedia.org/wiki/Cooperative_game_theory https://en.wikipedia.org/wiki/Shapley_value conversations with GPT-4 A…

Modified: September 12, 2023.

Tagged with: #game-theory

cooperative inverse reinforcement learning

References: Cooperative Inverse Reinforcement Learning The Off-Switch Game Incorrigibility in the CIRL Framework The CIRL setting models…

Modified: April 05, 2023.

Tagged with: #machine-learning#reinforcement-learning#alignment

cortex

Modified: February 25, 2022.

Tagged with:

counterfactual

Level 3 of Pearl's [ causal inference ] hierarchy: questions of the form 'given that (X, Y) happened, what would have happened if (X', Y…

Modified: August 02, 2021.

Tagged with: #causality

countries I've been to

USA (45/50 states: all but N Dakota, S Dakota, Minnesota, Lousiana, Mississippi) Canada Ecuador Costa Rica Argentina England Scotland France…

Modified: April 06, 2022.

Tagged with: #personal

creatine

In the cell, creatine is stored as phosphocreatine. It acts like a 'backup' adenosine: Phosphocreatine can can donate its [ phosphate…

Modified: January 19, 2022.

Tagged with: #biology

credit assignment

Modified: .

Tagged with: No Tags

crispr-cas9

"Crispr" == "Clustered regularly interspaced palindromic repeats". These are DNA sequences in bacteria that represent a genetic 'memory' of…

Modified: February 25, 2022.

Tagged with: #biology

cuda programming resources

easy introduction: https://developer.nvidia.com/blog/even-easier-introduction-cuda/ modern GPU (not so modern now): https://moderngpu…

Modified: .

Tagged with: No Tags

cumulative distribution function

Modified: .

Tagged with: No Tags

curiosity

Modified: .

Tagged with: No Tags

current identity, goals, and plans

This is an index of pages that reflect things I'd like to do, or that should otherwise should stay updated to reflect my current thinking…

Modified: September 11, 2021.

Tagged with: #personal

damn rockstar, always

Someone on tinder had this phrase in their bio. As a life motto I think it's pretty powerful. It's a short pneumonic for 'fuck the world…

Modified: March 02, 2022.

Tagged with:

data efficiency

Current (2021) deep networks require huge datasets in order to [ generalization|generalize ]. But we know that humans can do one-shot…

Modified: May 23, 2021.

Tagged with: #machine-learning

datacenter

How much compute does a typical Google data center have? Where would it fall on the supercomputer rankings? The top computer on the TOP50…

Modified: May 02, 2021.

Tagged with:

deceptive alignment

The idea is that a [ mesa optimizer|mesa-optimizing ] policy with access to sufficient information about the world (e.g., web search) might…

Modified: March 31, 2023.

Tagged with: #alignment

decision transformer

paper: Chen, Lu, et al. 2021, https://arxiv.org/abs/2106.01345 Trajectories are represented as sequences: where is the return-to-go, i.e…

Modified: April 15, 2022.

Tagged with: #ai#reinforcement-learning#papers

declarative and procedural knowledge

See also [ generative vs discriminative modeling ]

Modified: July 19, 2020.

Tagged with:

decoding

References: Holtzman et al. (2020), The Curious Case of Neural Text Degeneration https://arxiv.org/abs/1904.09751 How should we actually…

Modified: July 02, 2022.

Tagged with: #machine-learning

deconstructing sensory experience

A simple lens on meditative insight progress, from Michael Taft : Start with some sensory object - the sight of a tree, or the felt…

Modified: September 06, 2022.

Tagged with: #meditation

deep RL notes

Notes from John Schulman's Berkeley course on deep [ reinforcement learning ], Spring 2016. Value vs Policy-based learning Value-based…

Modified: February 22, 2022.

Tagged with: #machine-learning#ai#reinforcement-learning

deep deterministic policy gradient

Deep deterministic policy gradient (DDPG) is an interesting RL algorithm with a somewhat misleading name. Although its name indicates that…

Modified: July 27, 2022.

Tagged with: #reinforcement-learning

deep learning

see also: [ differentiable program ]

Modified: February 10, 2022.

Tagged with: #machine-learning

deep understanding

Which is more useful: reading the New York Times every day, or reading John Stuart Mill? listening to a current-events podcast, vs listening…

Modified: January 10, 2021.

Tagged with: #how-to-think

default mode network

A set of connected brain regions that are active when you're 'at rest', not focused on the external world. This includes mental states such…

Modified: February 25, 2022.

Tagged with: #meditation#drugs#neuroscience

defensive tech

Vitalik Buterin argues for defensive accelerationism (d/acc): One frame to think about the macro consequences of technology is to look at…

Modified: December 01, 2023.

Tagged with:

delayed sampling

Like quantum mechanics! We build up a distribution over variables defined so far. When we need to use a value, we sample from this…

Modified: May 16, 2020.

Tagged with: #papers#bayes

depression

Depression is the worst thing . Why? Ultimately we care about global utility. Depression is literally the state of finding it difficult or…

Modified: January 24, 2022.

Tagged with: #mental-health

dev tools

Google-internal equivalents: Dev tools: The ex-Googler guide (sourcegraph.com) GitHub - jhuangtw/xg2xg: by ex-googlers, for ex-googlers - a…

Modified: May 15, 2022.

Tagged with:

developing taste

The [ hedonic treadmill ] manifests in taste for things like wine, beer, coffee, fine cuisine. I've never spent effort refining my taste in…

Modified: July 24, 2020.

Tagged with: #growing-up#life-advice

differentiable environments

Maybe a stupid idea, but I wonder if the idea behind differentiable physics simulators (like Brax) can be extended more broadly to rich…

Modified: July 04, 2022.

Tagged with: #reinforcement-learning#ai

differentiable program

Fast differentiable sorting and ranking: https://arxiv.org/abs/2002.08871 What are differentiable analogues of 'standard' programming…

Modified: March 07, 2020.

Tagged with: #machine-learning

diffusion model

Diffusion models for image generation were independently invented at least twice: in a discrete-time variational inference framework…

Modified: August 31, 2022.

Tagged with: #machine-learning

diffusion process

References: http://www0.cs.ucl.ac.uk/staff/C.Archambeau/SDE_web/figs_files/ca07_RgIto_text.pdf https://www.ma.imperial.ac.uk/~pavl/lec_diff…

Modified: August 29, 2022.

Tagged with: #math

dihydromyricetin

Modified: .

Tagged with: No Tags

direct preference optimization

References: Direct Preference Optimization: Your Language Model is Secretly a Reward Model This seems like a compelling reframing of…

Modified: May 31, 2023.

Tagged with: #ai#reinforcement-learning

directions for probabilistic programming

Modified: .

Tagged with: No Tags

discount rate

Modified: May 27, 2021.

Tagged with: #finance

discrete latent variable

Modified: May 03, 2024.

Tagged with:

distinction

[ Otter notes ] August 2020: When somebody says that X is good---here X could be love relationships, money, peace, or whatever---it is never…

Modified: September 11, 2021.

Tagged with: #how-to-think

distributional RL

There are two forms of uncertainty in value-based [ reinforcement learning ]. Let be the return from trajectory , and be the expected…

Modified: August 04, 2022.

Tagged with:

diversification

Perhaps the only free lunch in [ finance ]. Given N investments all with the same expected value and level of risk (variance), whose…

Modified: May 15, 2021.

Tagged with: #finance

do-calculus

References: ML beyond Curve Fitting: An Intro to Causal Inference and do-Calculus , Causal Inference 2: Illustrating Interventions via a Toy…

Modified: August 06, 2021.

Tagged with: #machine-learning#causality

doing things yourself avoids cargo-culting

What does it mean to 'be yourself'? Selves aren't a well defined thing. Ultimately everything about your self is shaped by your surroundings…

Modified: May 01, 2023.

Tagged with: #life-advice

dopamine

Chemically, dopamine is a benzene ring (aka phenyl group), with two hydroxyl (OH) groups at adjacent sites, and a two-carbon (ethyl) chain…

Modified: December 21, 2022.

Tagged with: #biology#neuroscience

double descent

Empirically, as model capacity increases past the memorization threshold ( ), [ generalization|generalization ] error starts decreasing…

Modified: March 02, 2022.

Tagged with: #machine-learning

dream

Modified: .

Tagged with: No Tags

drugs

[ nootropics ] [ psychedelic ]

Modified: May 14, 2021.

Tagged with: #drugs

dual

Duality" is a deep concept in mathematics, but an intuitive way to think about it is in terms of tables. When we lay out data in a table, we…

Modified: September 09, 2024.

Tagged with: #math

dual gradient ascent

TODO: flesh out theory, understand ADMM (e.g., https://www.cis.upenn.edu/~cis515/ws-book-IIb.pdf )

Modified: July 06, 2022.

Tagged with: #math#machine-learning

dual metareasoning

From a conversation about [ attention ], [ multiplicative interaction ], and [ meta-reasoning ]: at some level, a lot of the AI problem…

Modified: October 17, 2022.

Tagged with:

dual-process cognition

Elephant and rider Asking "what should I value?" is asking the rider. it demands solving impossible moral questions. Asking "what do I value…

Modified: July 08, 2020.

Tagged with: #how-to-think#psychology

duality gap

Modified: .

Tagged with: No Tags

dukkha

In Pāli, dukkha refers to the unsatisfactory parts of existence. Literally it refers to a wheel in which the axel hole is not centered…

Modified: October 25, 2022.

Tagged with: #buddhism#mental-health

dullness

The fifth Buddhist precept is to refrain from intoxicants. Historically this means no alcohol. But a modern interpretation From [ Tucker…

Modified: .

Tagged with: No Tags

ego death

Modified: January 26, 2022.

Tagged with: #drugs#meditation

egregore

an idea that takes on a life of its own in the collective [ consciousness ] similar to Harari's concept of a shared myth. but Harari's myths…

Modified: .

Tagged with: No Tags

eight-fold path

Academic components: Right view (right understanding): seeing reality as it really is and understanding the [ four noble truths ]. This…

Modified: February 25, 2022.

Tagged with: #buddhism#life-advice

elastic

sources: wikipedia claude materials deform following a stress-strain curve. for many materials, this curve starts out approximately…

Modified: September 11, 2024.

Tagged with:

electronegative

Modified: May 14, 2021.

Tagged with: #chemistry

elephant path

Modified: October 29, 2021.

Tagged with:

eligibility trace

A few ways to think about eligibility traces: an explicit accounting of credit assignment a [ sufficient statistic ] for the history of the…

Modified: March 29, 2022.

Tagged with: #reinforcement-learning

embedded agent

Notes on Abram Demski and Scott Garrabrant's sequence on Embedded Agency Embedded Agents : Classic models of rational [ agency ], such as…

Modified: April 07, 2023.

Tagged with: #alignment#ai#buddhism

embodied cognition

Modified: .

Tagged with: No Tags

emergent capabilities

A consequence of [ phase transition ]s in [ large models ] is that models may end up having capabilities we didn't expect. For example…

Modified: April 07, 2022.

Tagged with: #ai

emotional labor

Relationships and community and mental health are not automatically maintained, and maintaining them doesn't come for free. Comforting…

Modified: July 17, 2020.

Tagged with: #relationships#growing-up#lgbt

emptiness

I think [ Dan Brown ] said somewhere that a good synonym for 'empty' in meditative contexts is 'mere construction'. For example, practicing…

Modified: March 23, 2023.

Tagged with: #meditation#buddhism

empty

See [ emptiness ].

Modified: March 23, 2023.

Tagged with: #meditation#buddhism

enabling environment

Andy Matuschak's concept of an Enabling Environment gets at something I've had in my mind but not named. It's an environment that expands…

Modified: January 24, 2022.

Tagged with:

energy-based model

Modified: February 10, 2022.

Tagged with: #machine-learning#modeling

enlightenment

David Chapman suggests that enlightenment in Buddhism is not a single defined thing, 'the word is hopelessly confused': https://vividness…

Modified: March 23, 2023.

Tagged with: #buddhism

enlightenment can't play chess

Very smart people tend to disbelieve in [ enlightenment ] because they hold up unrealistic notions of what it is or what it entails. There…

Modified: May 30, 2023.

Tagged with: #buddhism

ensemble

Often we think of ensembles in the context of supervised learning: we have some algorithm that learns X -> y mappings, and by running it…

Modified: February 10, 2022.

Tagged with: #machine-learning

entropy

Measures uncertainty, disorder, or randomness. The (Shannon) entropy of a probability distribution is: The quantity inside the…

Modified: April 15, 2022.

Tagged with: #math#bayes#physics

epinephrine

Modified: .

Tagged with: No Tags

epistemic uncertainty

Modified: .

Tagged with: No Tags

equanimity

Modified: .

Tagged with: No Tags

essays to reread

Innerring (C. S. Lewis). In any institution there are unofficial circles of influence: people who are 'on the inside' and people who are…

Modified: March 29, 2024.

Tagged with: #life-advice

ethanol

Probably the simplest and smallest molecule used as a psychoactive drug: It is neurotoxic, carcinogenic, and addictive, but of course…

Modified: August 01, 2023.

Tagged with: #chemistry#drugs

ether

The currency of [ Ethereum ]. Why does Ether have value? It represents computing time on a shared global computer. The more Ether you have…

Modified: October 08, 2021.

Tagged with: #finance

eudaemonic

Modified: January 25, 2022.

Tagged with: #morality

eurodollar

Foreign banks can create dollar-denominated liabilities much larger than their reserve of actual dollars, without the need to adhere to US…

Modified: March 13, 2022.

Tagged with: #finance