the word 'be' after modals

What this neuron does

This neuron reacts to the word 'be' when it comes right after a helper word like 'can', 'need to', 'going to' or 'please'. Turning it up made the model's writing wander and repeat rather than show any clear topic.

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

   -term, abstract goals that may have been set by their parents (“You will become a doctor!”), but clearly defined ambitions of students themselves. So, what can be achieved with great results in school? Forget long-term goals and focus on short-term

2. #2 peak

   a job seeker who cares about the long-term sustainability of the environment, there are many things you need to be doing right now that will position you to be able to participate in the development of

3. #3 peak

   heart or blood vessel problems, such as high cholesterol or high blood pressure, a healthy lifestyle can help you manage those problems.⏎If you have children, you can be their healthy role model. If your habits are healthy, your children are more likely

4. #4 peak

   proper application to view it first. In case of further problems read the IDEAS help page. Note that these files are not on the IDEAS site. Please be patient as the files may be large.⏎As the access to this document is restricted

5. #5 peak

   point.⏎Because many older people suffer from hazardous falls, having good vision is necessary to prevent them. Already having had a serious fall myself, I’m going to be very anxious to get this surgery done before my vision gets bad.⏎I urge everyone

6. #6 peak

   exercises or eating the wrong things.⏎To stop you from making mistakes at the dinner table, Bright Side has listed some of the most common diet mistakes. You could be guilty of committing some of these too!⏎1. Thinking that all calories are equal

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

   " uses 9 letters: A D E L M N O P S.⏎No direct anagrams for endoplasm found in this word list.⏎Words formed by adding one letter before or after endoplasm (in bold), or to adel

2. #2 peak

   ATRIP complex induces the Fanconi Anemia repair and Nucleotide Excision Repair (NER) pathways, as well as the p53 signaling pathway. C. Finally, a

3. #3 peak

   we don’t vote ordinary people are involved in political parties and in national/local government, they can be taken over by people who may represent a minority interest. People’s rights and freedoms are preserved by

4. #4 peak

   others can be purchased by visiting us at 111 Fairgrounds Road in Rolla, or by calling us. Low resolution images of many may be viewed by visiting the Geologic Map Index.⏎What is a geologic map?⏎Like all

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.