question mark after a rhetorical question

What this neuron does

This neuron lights up on the question mark at the end of a question an essay asks itself, like 'But what exactly is mulch?'. Turning it up made the model's output wander into loose, unfocused text, with no 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

   the practice isn’t effective we run the risk of making permanent misconceptions.⏎But what is “perfect” practice? Can “perfect” practice really exist? I do not think that “perfect”

2. #2 peak

   other, because both incur charges of “Denier!” to those who dare to question the consensus. Yet how solid is the consensus on human-caused climate change? Consider some recent news:⏎Angkor Wat’s Collapse From Climate Change Has Lessons for

3. #3 peak

   of Damascus,<|endoftext|>You’ve probably heard the term or maybe know someone who claims to be following such a food regimen: Clean Eating. What exactly does it mean? I read several articles, all with varying definitions and exceptions. I’ll do my

4. #4 peak

   question is whether the evidence is substantial enough to claim a discovery. And if a discovery can be claimed, the final question is what, exactly, has been discovered? Scientists can have complete confidence in their results but remain uncertain about how to interpret them

5. #5 peak

   doubled, to over 1.2 million between 2001 and 2011.⏎But how prevalent are highly diverse neighbourhoods across England? And should we expect their numbers to keep growing?⏎Multi-ethnic neighbourhoods in

6. #6 peak

   not the only thing requiring amends, and Indigenous people still face a great deal of social issues in modern day Canada. So what is the extent of these issues? And most importantly, what is being done about it?⏎Whilst only constituting

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

   and building was sanctified" [POOLE].⏎the tower of Meah--This word is improperly considered, in our version, as the name of a tower; it is the Hebrew word for "a hundred," so that the meaning is:

2. #2 peak

   “pumpkin pie” smoothie! Place the following in a blender:⏎• 3-4 tablespoons of canned organic pumpkin (just the pumpkin puree – because pumpkin pie filling has lots of added sugar)⏎• Two tablespoons of Greek yogurt

3. #3 peak

   ron sputtering is widely used to deposit thin films on a range of substrates, including glass and ceramics. The technology can be easily scaled to larger dimensions; it offers a high deposition rate (allowing a high level of productivity) and

4. #4 peak

   people.) For pints, use one tablespoon bottled lemon juice or ¼ teaspoon citric acid. Acid can be added directly to the jars before filling with product; in fact, this is recommended to be sure you get the acid in each and

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.