_index.org

Resilient Distributed Dataset

Last edited: August 8, 2025

Importantly, we have to keep our data under something that can be called RDD: “Resilient Distributed Dataset”; it is a theoretical dataset, but you don’t actually load it.

RDDs are has a single vector datastore under, but there are special RDDs that store key-value info. For Spark, RDDs are stored as operational graphs which is backtraced eventually during computational steps.

Pair RDD

A Pair RDD is an RDD that stores two pairs of vectors: you have a key and you have an value per entry.

resistor

Last edited: August 8, 2025

resistors do resistor stuff.

resistors in series

Their resistance add!

\begin{equation} R_{eq} = R_1 + R_2 + \dots \end{equation}

CURRENT remains the same through the resistors.

resistors in parallel

Their resistance add by inverse fraction!

\begin{equation} \frac{1}{R_{eq}} = \frac{1}{R_1} + \frac{1}{R_2} + \dots \end{equation}

VOLTAGE remains the same through the resistors.

restrict

Last edited: August 8, 2025

in C/C++, its a keyword:

f(restrict char *a,
  restrict char *b)

means that a and b are unique pointers to the objects to which they refer.

reticle

Last edited: August 8, 2025

A reticle is a photomask/template for a lithography system (like a negative). KLA was the first company to automatically inspect wafers and reticles.

reward model

Last edited: August 8, 2025

feed both accepted and rejected into your model, and get two scalars out \(r_{\text{rejected}}\), and \(r_{\text{chosen}}\):

\begin{equation} \mathcal{L}_{RM} = \log \qty(1 + e^{r_{\text{rejected}}-r_{\text{chosen}}}) \end{equation}

  1. train only for one epoch
  2. you should be getting low accuracy scores
  3. you may need to ensemble, margin loss

  • ppo gets the best model