profile pic
⌘ '
raccourcis clavier
Search

representations.

  • Étiquette

    • publié à

    • modifié à

    • durée

      2 min de lecture

    • source

      llms.txt

    See also: Edward Tuffe’s The Visual Display of Quantitative Information - ISBN-13: 978-1930824133

    Question

    How do we represent information flow when we interact with different mediums, esp. text?

    Maps shape how we see and understand the world, which in turn shapes how we act within it.

    maps

    Linus’ talk on Representation

    how it shape our agency

    as technology

    (Viégas & Wattenberg, 2023)

    How can we manipulate new ideas and abstractions to create new notations?

    Thesis: Technologies for representation mediate how we see the world.

    Tenet:

    1. A representation must abstract
    2. An interface always gives or takes agency. Be intentional about which you choose
    3. Represent to communicate, not model.

    internals grokking

    Though mechanistic interpretability to invoke agency

    Seeing clearly to exercise agency

    interface

    Instrumental use:

    • frictionless input, low-latency progress: payments.

    Engaged use:

    • see clearly from right perspectives. Express intent naturally and precisely.
    • Explore the possible spaces only through engaging with the information

    Why don’t we do this more?

    It is a hyperreal produced from a radiating synthesis of combinatory models in a hyperspace without atmosphere.

    Simulacra and Simulation, Jean Baudrillard

    Art of Cartography attained such Perfection that the map of a single Province occupied the entirely of a City […] saw that that vast Map was Useless...

    On Exatitude in Science, Jorge Luis Borges

    Bibliographie

    • Viégas, F., & Wattenberg, M. (2023). The System Model and the User Model: Exploring AI Dashboard Design. arXiv preprint arXiv:2305.02469 [arxiv]

    Vous pourriez aimer ce qui suit

  • Root locus and graphical analysis
  • DFAs, NFAs, and regular languages
  • Chaos is intuitive yet disheveled.
  • Zipf's Law
  • Min heap and binary search tree
  • Voltage-controlled electromechanical systems
  • ideas.
    • Aaron's notes
    Aaron's digital garden
    antilibrary.
    the one with all the books I should read.
    Garbage in Garbage out
    There has been this notion of “garbage in, garbage out” in CS which states that bad data, inputs will then produce an output that is of equal quality.
    Intelligence amplification
    I’m playing around with calling our tech, as it is today, IA (intelligence amplification) instead of AI. IA have the vibe of tools for thought, needing human interaction, and resemble a lot more what we actually have today.
    LLMs
    large language models, often implemented as autoregressive transformers models. GPTs and friends Most variants of LLMs are decoder-only (Radford et al., 2019) Have “capabilities” to understand natural language.
    Language
    Language as a public tool to understand the private life. important to our self-knowledge ⇒ emphasise through reading books. communication. Notably through the work of “Philosophical Investigations” by Ludwig Wittgenstein Concept of “language-game” The idea that each of us construct a pictures that we see the world through language.
    aesthetic value
    Also known as taste, under the scope of value Source whether artworks are necessarily aesthetic objects; how to square the allegedly perceptual basis of aesthetic judgments with ...
    autoencoders
    feature extraction from neural network
    latent space
    ...
    malleable software
    and the manifestation of the dynabook
    sparse crosscoders
    and how we observe multiple activation layers. SAE is a special case of sparse crosscoders.
    Gaussian elimination, LU decompositions, and errors LS solving
    Problem 1 [8 points] Consider the system Ax = b, where A=\begin{bmatrix} 0.1 & 0.3 & 0.9\\ 0.3 & 0.9 & 2.7\\ 0.6 & 0.7 & 0.1 \end{bmatrix} and b = \begin{bmatrix} ...
    Graphs
    See also slides Node as information and edges as relationship between data Directed acyclic graph (DAG) Application: Merkle DAG undirected. (\mathcal{N}, \mathcal{E}) \mathcal{N}: set of vertices \mathcal{E}: set of undirected edges: \mathcal{E} \subseteq \mathcal{N} \times \mathcal{N} path is a sequence of nodes and edges connect two nodes.
    Deterministic Finite Automata
    definition \Sigma^{*}: \text{set of all strings based off }\Sigma \begin{align*} \text{DFA}\quad M &= (Q, \Sigma, \delta, s, F) \\\ Q &: \text{finite set of states} \\\ ...
    Writing
    why writing? Writing as crystallised thought, a way of expressing the labyrinth of interconnected, messy, and incoherent ideas in my mind. It is a form of knowledge distillation (Jacky on writing) Writing is an exploration, an excavation of self and the world in this painfully intricate dance.