the word "be" after modals and "to"

What this neuron does

This neuron lights up on the word "be" when it follows words like "must", "need to", "has been proved to" or "to". It seems to mark a spot where a verb such as "be" is about to come, and it is quieter on the passive words that come just before "be" ("seen", "shown"). Turning it up did not produce a clear theme in the output.

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

   Appropriate Services in Health Care have taken into account the role of linguistic and cultural communication barriers in regards to patient-provider transactions.⏎Websites should therefore be properly translated to meet the needs of the Hispanic population. We need to consider cultural

2. #2 peak

   was found between dangerousness and withdrawal behavior mediated by negative affect, and between low ability and withdrawal behavior mediated by negative affect. The term used was not found to be associated with stigma. It is possible that previous stigma attached to the term, mental

3. #3 peak

   maybe Dropbox has an idiomatic way of writing Python with lots of "isinstance()"s.⏎> In general, we believe that pattern matching has been proved to be a useful and expressive tool in various modern languages. In particular, many aspects of

4. #4 peak

   Do the prep work in advance. Do as much prep work for a project as you can before you get the kids involved. Not everything on the page needs to be their work. It’s OK to help.⏎- Schedule several short sessions for one

5. #5 peak

   are carried over to the new data directory.⏎Stop the existing MySQL instance using the data directory. This must be a clean shutdown<|endoftext|>On April 16, 1853,

6. #6 peak

   2050 in a cost-effective way.⏎This study starts from the perspective that all gas consumption in Europe must, by 2050, be net zero carbon. This means that it

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

   way so that revolutionary businesses can do their thing.⏎This, of course, is hardly a novel view. Throughout the history of economic thought, government has long been seen as necessary but unproductive, a spender and regulator, rather than a value creator

2. #2 peak

   heaves debris (ejecta) outward from the point of impact. Regolith development on asteroids also follows the lunar pattern. On Mars, sand has been shown to make up a significant portion of the regolith, whereas on Saturn’s moon

3. #3 peak

   the world. Previously thought to be low in species diversity, the Antarctic is now known to have an unexpected richness of life. In addition, molecular techniques show Antarctica to be the birthplace of many species, driven by glacial cycles over millions of

4. #4 peak

   2017 also happens to be the target date for the first unmanned mission for NASA’s new Space Launch System

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.