'very' before an adjective

What this neuron does

This neuron fires on the word 'very' in phrases like 'a very small' or 'very little'. It also reacts negatively to digits buried inside numbers and reference codes. Turning it up gave some text that uses 'very cold', but also produced number loops, so the effect is only partly clear.

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

   link="none" columns="2" size="medium" ids="2568,2567,2569,2570"]These skills enable them to increase income, in

2. #2 peak

   speech services.⏎Other Service Providers⏎Except in the school setting, speech therapists can’t see patients without a doctor’s referral. Doctors have many patients to see and very little time. Just contacting them can cause conflict.⏎Nurses in the SNF

3. #3 peak

   light. When the beam is disrupted, an alarm will sound. It is most effective to warn about a smoldering fire. Ionization smoke detectors contain a very small amount of americium-241 within an ionization chamber. They

4. #4 peak

   or only negative are physically impossible, the international team came up with a way to do the next best thing. They created an effectively unipolar wave with a very sharp, high-amplitude positive peak flanked by two long, low-amplitude

5. #5 peak

   ⏎MRI scans can detect very small irregularities and produce a very high quality picture on its own, while the operator may also give the patient a

6. #6 peak

   the porous filter material is high-temperature alumina. It does not react with molten aluminum, does not dehydrate during the roasting process, has a very small shrinkage, and has relatively stable physical and chemical properties.⏎The slurry

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

   , MA: WGBH Media Library & Archives. Retrieved from http://openvault.wgbh.org/catalog/V_4E4EE3EAD24043E2BC0B972CD24B1

2. #2 peak

   405 Government. It means that it’s a 45 caliber bullet, pushed by 70 grains of black powder, and the bullet weighs 405 grains. Of course, things have changed, as we’ll see below

3. #3 peak

   is still pointing to the object created in aMethod().Instance variable. i value was changed to 20 in another(),so this prints 20. Thus final answer is 10 0 20. I

4. #4 peak

   astronomer, fellow skeptic, and fellow blogger Jeff Medkeff discovered the asteroid in 2000. It was given the preliminary designation of 2000 WG11, and Jeff had the privilege of naming it, and

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.