'New' in place names

What this neuron does

This neuron lights up on the word 'New' in place names like New York, New Orleans and New England, and on a few other place-name words such as 'Saint'. When it was turned up, the model drifted toward place-related words like 'metropolitan area', while turning it down pushed it into repeated phrases.

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

   depending on how much and how fast the state plans to cut pollution in coming years, a recent National Resources Defense Council analysis shows.⏎Maine and the five other New England states are among the nine states participating in the Regional Greenhouse Gas Initiative,

2. #2 peak

   . The bridge across Lake Pontchartrain into New Orleans is barely visible as a slightly brighter line against the lake's green and brown

3. #3 peak

   1, 1825, and construction completed on July 10, 1831. Since 1776 this portion of New York harbor has been a strategic base for coastal defense. Fort Lewis, once located

4. #4 peak

   burg and Iredell, and several mines, some still in operation, in Stanly county in the New London/Richfield area. But Gold Hill, just a few miles from Stan

5. #5 peak

   Ribaud, built Charlefort at Port Royal in 1562, and Fort Caroline under Laudonniere, at the River May (now St. John’s, Florida), in 1564. […]⏎Jac

6. #6 peak

   . His army passed through on York Road several times and Washington stopped over at the Hartsville Inn on several occasions, traveling between Philadelphia and Coryell's Ferry (New Hope). The action of the Battle of Crooked Billet took place just south

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

   headlines<|endoftext|>Impact of the introduction of the red-eared slider (Trachemys scripta elegans) on survival rates of the European pond turtle (Emys orbicularis)⏎- Cite this article as:⏎- Cadi, A

2. #2 peak

   would allow to be used as a concrete set accelerator⏎- No flocculation or coagulation⏎- All concretes and mortar containing steel reinforcement, post-tensioned concrete, concrete in contact with chlorides, parking structures, p

3. #3 peak

   (four-legged vertebrates) > Reptiliomorpha > Amniota >⏎Reptilia (reptiles) >⏎Romeriida > Diapsida > Arch

4. #4 peak

   eggs.⏎The sites anywhere the discoveries were made correspond to the upper Cretaceous, between the Campanian and Maastrichtian periods, some 70 to 83 million years ago.⏎The fossils found belong to small eggs

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.