mixed, mostly word fragments

What this neuron does

This neuron fires on pieces of longer words that are split in the middle, like 'tart' in 'tartar' or 'gravit' in 'gravitational'. It has no clear overall theme. Turning it up or down made the model write unrelated text.

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

   more sensitive and more likely to develop gum disease.⏎Inadequate oral hygiene routine: When you don’t brush or floss enough, plaque can turn to tartar. This buildup on and between your teeth can only be removed during

2. #2 peak

   tails, the researchers came up with an innovative robot design. They added a tail to their robot "Tailbot," which is essentially a toy car with a tail attached. It turns out that robot design could become much more than just a research tool

3. #3 peak

   Connecting⏎- 2 The Network⏎- 3 Printing Files⏎- 4 GuardBuddies⏎- 5 Radio⏎

4. #4 peak

   . Patton tromped into the room, ivory-handled pistol on his belt, jodhpurs flaring from his hips to knees. The general, tired and irritable from

5. #5 peak

   by Gizmodo have been flying for a month, the LIGO team wanted to be absolutely certain before making an official announcement.⏎That announcement has just come. Gravitational waves were observed on 14 September 2015, at

6. #6 peak

   and fatigue behavior.⏎Nanotubes generally have diameters ranging from a few nanometers to a few hundred nanometers. Because of their strength, carbon nanotubes can be used for mechanical structures and biomedical applications.⏎In the previous project,

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

   1.2 and x = (9/10)log102 ≈ 0.97. The expected number of powers of 2 without zeroes starting from the power N is cxN/(1-x) ≈

2. #2 peak

   one side of an equilateral triangle and the circle touches the other two sides of the triangle. A smaller circle touches the larger circle and. . . .⏎X is a moveable point on

3. #3 peak

   nom Da, Takeo is about 102 km away from Phnom Penh. It is about 24 km away from the town of Takeo.

4. #4 peak

   .1 µF each, non-polarized (Radio Shack catalog # 272-135)⏎- Two 27 kΩ resistors⏎I recommend ceramic disk capacitors, because they are insensitive to polarity (non

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.