'by' before a deadline

What this neuron does

This neuron fires on the word 'by' when it sets a deadline, as in 'by 2050' or 'gone by summer'. Turning it down made the model repeat the word 'now' over and over, which fits the neuron pushing against that word.

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

   ⏎Cities are crucial to achieving the SDGs: more than half of the world’s population lives in cities, with this number predicted to increase to over two-thirds by 2050. Mapping out an implementation strategy for SDG 1

2. #2 peak

   well as a similar study did find that cats decl<|endoftext|>With the human population expected to grow by billions over the new few decades, the United Nations is predicting that by 2050 our food production must double

3. #3 peak

   measured in the metric tonne that are verifiable and tradeable. Bloomberg predicts that the price of carbon credits may rise by 3,000 percent by 2029 with better regulations.⏎The ever increasing demand for carbon credits

4. #4 peak

   not be surpassed by the end of the century. That level, they say, is 2 metres: sea levels are unlikely to rise more than 2 metres by 2100.⏎Setting an upper limit on how fast seas will rise

5. #5 peak

   future legislative action and planning.⏎Research included in the report suggests Hawaii will see nearly a one-meter rise in ocean levels by the end of this century. It predicts that more than 6,00

6. #6 peak

   ers Final Test Phase⏎The COVID-19 pandemic seemingly has no expiration date. At its onset, we thought there was a decent chance it would be gone by summer. Fast forward and we see more cases in the summer than we ever did

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

   half.⏎- Starting and ending at the fold, make the first stroke of the number 2.⏎- Cut along the line.⏎- Unfold the paper.⏎- Use the same method to make more hearts. Use construction paper, gift wrapping paper

2. #2 peak

   . First the cardinal numbers. That's only one 3. This will remain. And then the numerators. 1 + 3 = 4. We then get the answer 345 Can we simplify? No, that's

3. #3 peak

   up to 1.⏎- Now select the ball, and go to the Physics tab. Also click ‘Rigid Body’ and leave the type as ‘Active’. We want this to simulate animation. Now change the bounciness to 1

4. #4 peak

   right, 256 units up). Now rotate it 90 degrees. We have now created the beginning and end of the 90 degree pipe.⏎Now, lets create the splines to

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.