opening parenthesis after a noun

What this neuron does

This neuron fires on an opening bracket "(" that comes straight after a noun, where the text is about to add a short aside such as a definition, an abbreviation or an example. Turning it up made the model drift into loose text with brackets, but no clear theme came out.

Written by Claude from this neuron's activations and steering outputs. Confidence: medium. Think it is wrong? Flag it (opens a prefilled issue on GitHub).

Fires most on

The text windows where this neuron's activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

1. #1 peak

   the Terrible till this day, Russia has been trying to present itself to the Western power as the perfect economic trading partner. Moscow insists that “small states” (Poland, Ukraine, etc.) are basically the

2. #2 peak

   .⏎A post-cubist squarist artwork is painted on a rectangular grid, in which each grid cell is either coloured or blank. Two adjacent coloured cells (vertically or horizontally, not diagonally) are connected, and a shape is

3. #3 peak

   to prevent transmission of infectious agents from one person to another. Needles should be disposed of properly and not recapped.⏎It is recommended that small syringes (0.5 mL or 1 mL) should be used to minimize any product

4. #4 peak

   Cosway, Rhys’ take on Bertha Mason, the mad wife of Mr. Rochester in Brontë's classic. Although now hailed as a masterpiece (TIME magazine named it among the 100 best English-language novels since

5. #5 peak

   materials is creating big data.⏎Take, for example, the emergence of Moocs – massive open online courses. While some universities (and indeed some individual faculty) are offering Moocs, new entrants to the

6. #6 peak

   the sinewy chromatic theme, the foreboding against the triumphant. At the very outset, Elgar pens one of his most stately melodies (he already had four of the five “Pomp and Circumstance” marches

Steering

Push this neuron up or down while the model writes, and watch what changes. The outputs were generated ahead of time, so nothing runs in your browser.

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

1. #1 peak

   Col. Stand Watie, with one regiment of Cherokees, to move into the neutral land

2. #2 peak

   conuts and banana leaves<|endoftext|>const void * memchr ( const void * ptr, int value, size_t num );⏎void * memchr

3. #3 peak

   . Jose Rizal made patriotic annotations on Morga<|endoftext|>Early in the 14th century the Turkish tribal chieftain Othman (Osman) founded an empire in western Anatolia (Asia Minor) that was to endure

4. #4 peak

   raspberry', 'pomegranite', 'bananana', 'grape', 'loganberry'] print "You are an" , random.choice (verb) , random.choice (fruit)⏎Any Tutorial Videos?<|endoftext|>What is global

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Activation histogram

How often this neuron fires at each strength. Almost always it does close to nothing (the tall bars in the middle); the snippets above are the rare moments it fired hard, amber for positive and blue for negative.

Logit lens

The words this neuron nudges the model toward (amber) or away from (blue) when it fires. A hint, not a label: many neurons work through later layers instead.