'of' after 'think' or 'out'
A neuronOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in layerOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. 0 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featuredA neuron picked for the full treatment: steering outputs and a written explanation. About 3,000 of the 86,000 neurons are featured., so it has steering outputs.
What this neuron does
This neuron lights up on the word 'of' when it comes right after phrases like 'Think of' or 'way out of'. Turning it up did not produce any obvious change in the writing's theme.
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 activationThe number a neuron produces for a token. Large positive means it fired hard; negative means it pushed the other way. was highest. Amber marks positive activation; the outlined tokenA piece of text the model reads at once: often a word, sometimes part of a word or a punctuation mark. is the peak. Hover a token for its strength.
-
're looking to help the bees in your hood, consider adding some native flowering plants to your garden.⏎"Think of the flowers your grandmother used in her garden as a practical guide, especially when using
-
and use in sentences; and questions for both literary analysis and critical reading. Each week the student is challenged to "make it real." For instance: "Think of two examples of foils in literature or movies" or "Name at least one
-
write a setting description.⏎Below is a picture of the outside appearance of the lighthouse. There is also a picture of the inside of a lighthouse. Think of interesting words to describe the looks and feel of the lighthouse. If you can
-
a success?⏎19. It can’t be spring if your heart is filled with past failures.⏎20. Don’t think of it as failure. Think of it as time-released success.⏎21. There are no failures. Just
-
memories with a computer. An Imagination Interface, perhaps. By any name, it doesn’t take much time to realize that you have always wanted one. Think of a world where you could selectively download your life’s memories- sights, sounds,
-
_t;⏎This type is used to define difference between two pointers. Depending on which of the pointers is larger, this value may be positive or negative.<|endoftext|>Think of your body like a car. Where the muscles are the shock systems and the skeleton
SteeringAdding a push along a neuron while the model writes, to see how its output changes. Turning it up or down.
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.
Here is something I have been thinking about lately: I am thinking about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about about
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
health workers are battling to save lives in the midst of shortages of personal protective equipment, oxygen, and treatments.⏎“Vaccines have never been the way out of this crisis on their own, but this current wave is demonstrating again just what a
-
planetary security forces. Almost every smugglers vessel had improved light speed and sub-light speed drives for increased speed as well as boosted weapons systems to fight their way out of tough situations. Another of the most prevalent modifications to a smugglers freighter were
-
member’s home or the Women's Center shelter.⏎• Keep the Women's Center toll-free hotline phone number close at hand.⏎• Figure out the quickest way out of your home and practice the escape.⏎• Obtain an Order of Protection.⏎• Pack
-
scope, abundance, it is because he has inherited the achievements of his forerunners. Nothing warrants the idea that he will ever hew his way out of the half-knowledge half-ignorance which is the stamp of his kind
Connections
Show on mapWhich neuronsOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in earlier layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.
Fed by earlier layers
None: this is the first layer.
Feeds later layers
- L18:778 excites
- L22:1031 weakly excites
- L22:2705 weakly excites
- L23:2959 weakly suppresses
- L26:1089 weakly excites
- L20:3008 weakly excites
- L20:392 weakly excites
- L1:420 weakly suppresses
- L19:595 weakly suppresses
- L18:110 weakly excites
- L21:1060 weakly excites
- L20:1962 weakly suppresses
- L8:857 weakly excites
- L23:309 weakly excites
- L12:2771 weakly suppresses
- L20:2681 weakly excites
- L6:346 weakly suppresses
- L6:1782 weakly excites 'of' after a category noun
- L22:384 weakly excites
- L2:3052 weakly excites
- L20:2414 weakly excites
- L27:797 weakly excites
- L27:2806 weakly excites
- L12:2053 weakly suppresses
Activation histogramA chart of how often a neuron takes each activation value.
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 lensReading which output tokens a neuron pushes toward or away from, straight off its output weights, before the later layers have their say.
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 layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. instead.
Promotes
- yssey
- emailer
- Observatory
- INTR
- umo
- ample
- lico
- PRS
- transform
- ROUP
Suppresses
- benh
- /pp
- Arbor
- ッシュ
- rear
- usted
- 建成
- illation
- 螂
- sizes