comma after sentence-opening 'Oh' or 'So'

What this neuron does

This neuron responds to the comma (or exclamation mark) right after a short opener such as 'Oh', 'Um', 'So' or 'Or' at the start of a clause. Turning it up made the model write in a loose, repetitive style, and turning it down made the text more formal.

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

   and hippos. Some parents unparent, others choose to follow French mama’s and papa’s and let’s not leave out the free rangers. Um, does it seem like there’s a new style added to the list almost every day

2. #2 peak

   value⏎> associated with key represented by the leaf be (re-)inserted into the⏎> Map. The Tree itself need not change at all.⏎Oh, it will. If you change a Map, you get a new one, thus

3. #3 peak

   Indeed,⏎Constitutional experts say the motion carries no legal weight and would be unlikely to give Quebec separatists a lever to extract more political powers from Ottawa.So, indeed, this would be a non-nation, nation. In fact, what

4. #4 peak

   But in situations where the names of your Entities, Structures, and Attributes are long, this will save you a few keystrokes and is easier to read.⏎Oh, and you’ll notice that we’ve also updated the defaults. So, if you

5. #5 peak

   . I don't, besides that I was in residency.⏎I bet you have no idea how close you came to dying (it was 2 minutes). Or,

6. #6 peak

   the river Sabine, in the sea, continuing north, along the western bank of that river, to the 32d degree of latitude; thence, by a line due north, to the degree of latitude where it strikes the Rio

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

   in PNAS⏎Our ideas of protein function have undergone a marked transformation with the discovery of an aspect of enzyme evolution called "moonlighting." That is, a protein can acquire a second (moonlighting) function without concomitantly

2. #2 peak

   yl chain, or a short stretch of ethylene glycol ethers, up to as complex a linker as you have the patience to search for.⏎It seems like a rather simple idea that you could get two different sorts of activity from the same molecule

3. #3 peak

   the new electricity” (Artificial intelligence is the new electricity). Meaning, AI affects every aspect of economic activity and almost every task in our daily lives. So much so that a country lagging behind in AI issues actually runs the risk of being

4. #4 peak

   and trans-Caribbean approach to investigate the different strategies developed by elite men of color after the Haitian revolution to secure their status in the Atlantic World. Considered as “the first truly modern international crisis of exiles,” the

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.