unclear: word fragments and 'or'/'and' in lists

What this neuron does

The pattern is hard to pin down. It tends to fire on pieces of split words, such as the middle of a name, or on words like 'and' and 'or' inside lists. Turning it up made the model drift toward repeated plain phrases like 'the weather outside was fine', with no clear theme.

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

   Stars⏎An accessible introduction to neutron stars that includes details on their life cycles as well as detailed information on neutron star variants such as x-ray and gamma ray bursters.⏎Introduction to Pulsar Astronomy⏎A tutorial on radio

2. #2 peak

   Mark⏎A. Mandel, Charles Häberl, Will Bennett, Johannes Reese, Cristina Matute⏎The following is

3. #3 peak

   best to utilize other forages during this period of growth, such as annual warm-season grasses or legumes where possible. You might also consider feeding hay or grain.⏎De-stock pastures. Livestock pressure on pastures can be

4. #4 peak

   gets its name from its size — it is the largest of all North American rails, a group of marsh-dwelling birds that includes moorhens and gallinules.⏎It’s a year-round resident of Florida and found throughout the state

5. #5 peak

   fail, find on the wall next to the art, a statement from the artist about their intent for the piece.⏎When Marcel Duchamp submitted an upside-down porcelain urinal signed “R.Mutt”<|endoftext|>The Galileo Museum in Florence is an

6. #6 peak

   Sons first. As far as I can tell, they both sprang from the same War to End all Wars, against the Greek King Antiochus Epiphanes who decided to wipe the Jewish nation from the face of the earth in

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

   from Science Advances, it was believed there were two species of of wolves in the United States: gray wolves (Canis lupus) and red wolves (Canis rufus). Gray wolves were brought in from Canada to repopulate Yellowstone

2. #2 peak

   igure of splicing mechanisms)⏎Splicing of nuclear mRNA is catalyzed by specialized RNA-protein complexes called small nuclear ribonucleoprotein⏎particles (snRNPs). The RNAs found in snRNPs are identified as U1

3. #3 peak

   digital records for use in K-12 lesson plans, visit the IHRC’s portal for digital resources via the University of Minnesota’s UMedia Archive:http://ihrc.umn.edu/research/digitalsources.php⏎Click here for additional

4. #4 peak

   points in Scrabble, and 8 points in Words with Friends⏎There are 5 letters in rifts: F I R S T⏎2 Anagrams of Rifts⏎Words that can

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.