unclear: split words and dimensions

What this neuron does

This neuron fires on a piece of a word that is cut into several parts, like 'trifoliate' or 'foliaceous', and on words that describe directions or shapes, like 'vertical' or 'leaves'. It goes strongly negative on measurements such as '3 x 3 x 3' or 'm2'. Turning it up didn't produce a clear theme in the writing.

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

   twiggy bush, of fully a yard high. Leaves trifoliolate, clothed beneath with closely adpressed hairs, and bright yellow, somewhat

2. #2 peak

   - blue paper⏎- black paper⏎- pencil top eraser⏎The piece we selected to recreate was Van Gogh’s Starry Night. We’ve been learning about apples this week, an easy tie in was to

3. #3 peak

   UN committee asked the Australian Government to report back to it in detail in the<|endoftext|>Discover the words hidden in the grid, written through one another in eight horizontal, vertical and diagonal directions. Select words by clicking on the first and last letters, in

4. #4 peak

   first person to cover 4 pictures in a row (horizontal, vertical or diagonal). The winner calls out "Lotto!" or "Loteria

5. #5 peak

   level as below.⏎Size of QRcode is defined as version.⏎Version is from 1 to 40.⏎Version 1 is 21*21 matrix. And 4 modules increases whenever 1 version increases.⏎So version

6. #6 peak

   short lateral stems and then seemingly axillary)||Muellerina|⏎Back to 1⏎|3||Flowers in heads, enclosed in 2 enlarged foliaceous bracts fused at margins over buds during development||Diplatia|⏎

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

   be loaded into your 3D printer. If it’s your first time using a 3D printer, consider starting with something relatively small (3” x 3” x 3”) so you can get comfortable with the 3D

2. #2 peak

   or effective dilution and removal of the gas. A release scenario showing a 10 g/s release of hydrogen inside a process container of 40 m2 is illustrated in figure 4. The calculations, carried out with the CFD

3. #3 peak

   . When students recognize that they need the lengths of the diagonal sides of the box give them the measurements for those sides (2.2 inches for the sides around the top and 6.4 inches for the sides

4. #4 peak

   of rock, usually about 0.03 mm thick, attached to a glass slide with an epoxy resin. Thin sections are usually about 26 x 46 mm, but there can be other sizes depending on what needs to be

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.