words before a 'suggests' style claim

What this neuron does

The neuron fires on the word that sets up a conclusion drawn from evidence, like 'This strongly suggests' or 'would suggest'. Turning it up made the model repeat the word 'suggested'.

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

   each flagellum faces the other flagellum, indicating that each axoneme has the same rotational orientation relative to the direction of its effective stroke. This strongly suggests that the direction of the effective stroke is controlled by a structural component within the

2. #2 peak

   in a field in Aberdeenshire.⏎The new finds include late Roman coins, pieces of late Roman silver vessels, bracelet and brooch fragments and other objects that would have been highly prized objects.⏎Dr Gordon Noble, Senior Lecturer, Department

3. #3 peak

   end fields of study. There is one key question you should be asking yourself: Is the replication crisis evenly distributed across different scientific disciplines? My reading of the literature would suggest “no”, but the reasons

4. #4 peak

   .⏎BAR coauthor Lily Singer-Avitz suggests that several finds discovered in the later strata, including Egyptian-style seals and seal impressions and local pottery sherds, should be associated with a Late Bronze Age–Early Iron I period presence at Tell

5. #5 peak

   dark patches do not reflect radar beams back to the probe, suggesting they are uniform, flat surfaces. The probe has also found dark, branching features on the moon that appear to be river systems.⏎"You have to come up with some really pathological material

6. #6 peak

   scholarworks.uni.edu/pias/vol31/iss1/25<|endoftext|>New images from NASA’s Mars Reconnaissance Orbiter show finger-like markings that stream down Martian slopes when temperatures change, which would most likely occur if liquid water

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

   new experimental access to quantum physics has been made at University of Konstanz. The team of scientists headed by...⏎Yersiniae cause severe intestinal infections. Studies using Yersinia pseudotuberculosis as a model organism aim to elucidate

2. #2 peak

   same. In a crash, a person not buckled up is much more likely to be injured or killed than someone wearing a seat belt.⏎- Never speed – Research done for Ford Driving Skills for Life shows that if parents speed, their teens are

3. #3 peak

   . But Ratto and coons aren’t simply putting their designs, software and tutorials online and calling it a day: Just because something is available in theory doesn’t automatically make it accessible to all. (Linux, anyone?)⏎Like a whole cadre

4. #4 peak

   achievement underscores the nutritional challenges posed by an expanding world population. "Earlier this year, the Food and Agriculture Organization of the United Nations (FAO) released a report titled, 'Edible Insects: Future Prospects for Food and Feed Security

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.