'once' after was/were/have

What this neuron does

The neuron fires on the word 'once' when it comes after a verb like 'was', 'were' or 'have' and means 'at some time in the past'. Turning it up made the model write more about past events and history.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

   an Emperor/ress before the Shoguns took complete power. Samurai probably existed well before the appearance of Shoguns.⏎It is also known that there was once an ancient civilisation in the area pre<|endoftext|>- Herbert A. Simon was an

2. #2 peak

   , I get excited about this.)<|endoftext|>It is supremely closed-minded to throw away Homer and Mozart and Augustine.⏎Professor, music historian and composer Robert Greenberg once received a letter from a musician urging him to give up the traditional and classical repertoire

3. #3 peak

   the labor movement and I highly doubt many children even know why they are out of school. Many have no idea children were once forced into backbreaking jobs in mines, textiles, glass factories, canneries

4. #4 peak

   instructions to his eldest son, Louis X⏎If God send thee adversity, receive it in patience and give<|endoftext|>In a remote part of the Olympic Peninsula there was once a tiny seaport community

5. #5 peak

   thagoras travelled east beyond Egypt and came back with the notion (but they say this of Jesus too).] A later report claims he told followers that he has once been Aethalides,

6. #6 peak

   “Bookends” couldn’t describe the decade of freedom and love better. What a time it was! Paul Kantner of “Jefferson Airplane” was once quoted as saying “If you can remember anything about the 60’s,

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

   having the encrypted string is pointless. However, this does NOT provide any meaningful security against a determined attacker as it is rather trivial to open up the executable in a debugger and simply look at the decrypted information as the program runs. This is only meant

2. #2 peak

   for validating expression of miRNAs in whole genome analyses such as microarray studies7. The samples used in this protocol were from patients who underwent radical prostatectomy for clinically localized prostate cancer; however other tissues and cell lines can be substituted in

3. #3 peak

   million people and produced economic losses of more than $1 trillion. But the agency pointed out that efforts to better deal with disasters have begun to bear fruit: even as the number of disasters increased 4-fold, the number of fatalities decreased by nearly

4. #4 peak

   panther’s last remaining core habitat area. It would facilitate development that would bring 300,000 additional residents into this rural region.⏎Even as the federal agency acknowledges the significance of this proposal,

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.