"to" after "used" (purpose)

What this neuron does

This neuron lights up on the word "to" when it comes right after "used" and introduces a purpose, as in "used to treat" or "used to analyze". Turning it up made the model's writing collapse into loops of the word "use", so the link to the text it writes is weak.

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

   in electrical current—can plunge the results of the most carefully designed experiment into ambiguity.⏎Consider the laser, one of the workhorse tools of modern research, used to analyze, print, scan, cool, heat, and more. Variations in

2. #2 peak

   Carpobrotus virescens). The fruit provided a sweet, thirst quenching, summer food and an infusion of the crushed leaves was used to treat stomach cramps and diarrhoea.⏎Quandong (Santalum

3. #3 peak

   found in kitchens all around the world. But did you know that black pepper is not only the world’s most traded spice, but that it can also be used to illustrate some important science principles such as surface tension?⏎Here’s a fascinating home science

4. #4 peak

   theft. It can be both a state crime and a federal crime.⏎- Credit Card Fraud⏎When a credit card is stolen, forged or in any way used to pay for something without the permission of the card owner, it is credit card fraud

5. #5 peak

   flower, or fruit from which they are derived. Some, like Lavender and rose, are used to soothe and calm. Other essential oils are more invigorating — basil and

6. #6 peak

   iterpenes, phenolic compounds, carotenoids, steroids, and ketones. Its branches and leaves are often used to produce essential oils and metabolic extracts employing different organic solvents (Ferreira et

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

   their citizens of ethnicity Y who have contacts in State B are supporters of State B, and try to identify, capture and interrogate them to have information that may be helpful to suppress any opposition.⏎[F4] Security services of State A hack into

2. #2 peak

   will cut pollution, and that decision could have a significant impact on the money each state earns under RGGI pollution permit auctions. That money earned in turn can be directed toward clean energy programs to help industry and residents cut their energy bills.⏎R

3. #3 peak

   to start a dialogue about racial representation and stereotypes.⏎The program follows student participation in an education partnership with the West Chicago City Museum. All proceeds from the performance will be donated to the school’s Multicultural Sensitivity Club.⏎Thoughts on the program

4. #4 peak

   a broad range of cancer cells. It enables the cancer cell to consume far more glucose than normal, while using little of it for energy. Instead, the rest is used to<|endoftext|>The turn of the twentieth century is often referred to as the nad

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.