unclear: answer labels and closing quotes

What this neuron does

This neuron is unclear. It fires on things like '(a' at the start of answer options and the number 2 in a formula, and it seems tied to what comes next in a list or quote. Turning it up gave a mix of unrelated text, so the pattern is not clear.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

   10 log10(MAX/(MSE)^(1/2))^2 = 20 log10(MAX/(MSE)^(1/2)). Therefore, PSNR = <|endoftext|>In my current position as a veterinarian working

2. #2 peak

   (usually Thomas of Cana) with the Thomas; but it is undeniable that the Nasrani or

3. #3 peak

   Express MSE mathematically from the description in Step 1. We therefore have MSE = 1/mn [?? (I(i,j) - K(i,j))^2] where I and K are matrices that represent the images being compared.

4. #4 peak

   illuminating being the incandescent electric bulb, the phonograph and the motion-picture projector. Who was he?⏎- Match the following with their inventors:⏎(a

5. #5 peak

   invent in 1800?⏎- What was invented solely for the use of King Louis XIV of France?⏎- Who invented these articles of everyday use:⏎(a) Safety razor⏎(b) Ironing board⏎(c) Friction Matches⏎

6. #6 peak

   .⏎Printer’s Ink — 1718⏎Ben sold 830 newspapers. His brother James sold 177 fewer newspapers than Ben.⏎(a) How many newspapers did James sell?⏎(b) How many newspapers did they sell

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

   providing the assurances ("how did we get here?") that led to action (including the action itself) while the introducer chain ("where do we go from here?") provides the assurances both for a continuation of that action and for other actions

2. #2 peak

   320 ft (6,194 m) Mount McKinley or Denali (its Koyukon Athabaskan name meaning “The Great One”). The majestic mountain, which has extremely cold weather with temperatures as low as -

3. #3 peak

   business outcomes are too coarse-grained to be used to directly write software (where do you start coding when the outcome is “save 5% of my operating costs”?) so we need to define requirements at some intermediate level in order to get

4. #4 peak

   an ex-partner⏎Many other respondents were also doing things which may prevent them from building up a good credit history: for example, one in ten (11

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.