"and" inside fixed institution names

What this neuron does

This neuron lights up on the word "and" in set phrases like "National Oceanic and Atmospheric Administration" or "National Institute of Allergy and Infectious Diseases", where the next word is almost certain. When it was turned up, the model's writing did not visibly change in any clear way.

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

   by the US National Oceanic and Atmospheric Administration (NOAA) (NOAA, 2020).

2. #2 peak

   - Ensure that your animals’ needs for water and shade are met.⏎What should I do during a heat wave?⏎- Listen to a NOAA (National Oceanic and Atmospheric Administration) Weather Radio for critical updates from the National Weather Service (NWS

3. #3 peak

   made it possible to grow marine finfish in the open ocean sustainably according to a new report released by the National Oceanic and Atmospheric Administration (NOAA).⏎Specific types of fish farming can be done with minimal

4. #4 peak

   Thomas and the U. S. Virgin Islands. Numerous reef animals are being protected from pollution and harvest for aquarium sale. (Department of Commerce and National Oceanic and Atmospheric Administration, 1999; Department

5. #5 peak

   to give early warnings of events. As part of this, we are working in collaboration with the US’s National Oceanic and Atmospheric Administration (NOAA) to develop<|endoftext|>The formation of counties was one of

6. #6 peak

   some in the Bay Area are getting more housing. Two marine sanctuaries off the coast of Northern California will more than double in size, the National Oceanic and Atmospheric Administration has ruled.⏎Florida might not like to talk about climate change, but

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

   cardinal numbers and the numerators. First the cardinal numbers. That is 2 + 3 = 5. And then the numerators. 1 + 1 = 2. We then get the answer 524 Can we

2. #2 peak

   . Regulator models vary if steam pressure at regulator is less than or equal to 15 PSIG or greater than 15 PSIG but less than or equal to 125 PSIG.⏎- Regulator Size not applicable.

3. #3 peak

   Robert E. Stevens, GENESIS II (The Second Beginning) E-Mail: firstname.lastname@example.org⏎I am not talking about truth in that 1 + 1 = 2, but

4. #4 peak

   noise, altitude, navigation, and maneuverability readings. Then you hard-wire the devices into place, and send the Test Pilot into the clouds. When he or she returns, you collect and analyze the information gathered by the instruments you designed.

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.