'sh' starting a word piece

What this neuron does

This neuron fires on the 'sh' sound at the start of a split word, as in 'shatter', 'shins', 'shelve' and 'shoveling'. When it was pushed up, the model fell into repeated symbols such as asterisks or line breaks, and when pushed down it produced rows of blanks.

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

   from deep pumped bores in vertical transmissive fault or shatter zones could augment mountain rivers for power generation. Flows might be augmented in

2. #2 peak

   Implicit and Explicit Processing⏎We often learn a task by telling ourselves what to do. In the squat, I think about knees tracking out, butt going back, shins perpendicular, and opening up the hips on the way down. The explicit directions

3. #3 peak

   to struggle to make reasons for not doing things they obviously ought to do. When the particular thing they ought to do involves the encouragement of art, their struggles to shelve the whole matter become quite painful. As an illustration of this, the case

4. #4 peak

   million tons of plastic bottles. Plastic water bottles offer many advantages to both the manufacturer and the consumer. It is lightweight, which lowers transportation costs. It does not shatter, making it a safer option. However,

5. #5 peak

   tibia) hurts when you press on it firmly with your finger.⏎The causes of shins splints include problems with body mechanics, a sudden or excessive increase in the

6. #6 peak

   bi-products such as saw dust and planer shavings. This production demanded large amounts of water, and thus the pump tower and

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

   and regulate blood pressure.⏎- Hijiki has more calcium than any other sea vegetable (14 times more calcium than milk), and is high in fiber.⏎- Nori contains as much protein as soybeans and

2. #2 peak

   Описание презентации Презентация london-history-12162149479

3. #3 peak

   flows, dotted: scoria cones. // in Elliot Hildner / University of Bremen⏎The main crack<|endoftext|>analysis of algoritms⏎robert.kern at gmail.com⏎Fri Sep 10 00:10

4. #4 peak

   an ancient temple for Lord Krishna. Dwapara Yug<|endoftext|>Simple technique appears to be safe and effective, review suggests⏎By Robert Preidt⏎MONDAY, Oct. 15 (HealthDay News)

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.