mixed, numbers and list items
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. 16 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. 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 fires on a scattered mix of words and digits, such as a digit inside a number range, a word in a list of topics, or a part of a longer word. No single theme stands out. Turning it up made the model repeat itself, so it does not reveal much on its own.
Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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.
-
6. He got very depressed after his failure.⏎7. Jessica’s failure was due to her idleness.⏎8. Jessica’s failure was due to her idleness.⏎9
-
That is, it would be discounted from the final price.⏎- How to calculate percentage in Excel⏎- Find the difference between two numbers⏎- Calculate percentage in Excel based on two quantities⏎- How to Calculate Excel Percentage in Multiple Cells⏎-
-
30⏎- Roman Numerals 1-50⏎- Roman Numerals 1-100⏎- Roman Numerals 1-200⏎- Roman Numerals 1-500⏎- Roman Numer
-
55% humidity to manage health effects and illnesses.⏎Comfortable: 30% – 60%⏎Recommended: 45% – 55%⏎High: 55% – 80%⏎Contact Air Care
-
our numbers, but the injury to Jewish...⏎Become a Member⏎Read all member content⏎Get exclusive in-depth reports from Israel.⏎Get exclusive in-depth reports from Israel⏎Connect with Israel, right from your home.⏎Lift up the voice of
-
.⏎For thirty winters Theodric ruled⏎the Mćring stronghold with an iron hand;⏎many acknowledged his mastery and moaned.⏎That passed away; this also may.⏎We have heard too of Ermanaric's wolfish ways,⏎
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 have been thinking about lately... I have been thinking about lately... I have been thinking about lately... I have been thinking about lately... I have been thinking about lately... I have been thinking about lately... I have been thinking about lately
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
each with one of the following statements written in large letters: “Strongly Agree,” “Agree Somewhat,” “Disagree Somewhat,” “Strongly Disagree”⏎Activities / Procedures:⏎1. WARM-UP/DO-NOW
-
ays differential relays, and distance relays.⏎According to timing characteristics the relays can be divided into instantaneous, definite time lag, inverse time lag and inverse definite minimum time (IDMT) lag relays.⏎In induction type of relays
-
of these tenses has four sub-categories (or aspects) which better describe the tense used within the writing. These aspects are simple, perfect, progressive, perfect progressive. Confused yet?⏎Simple Past: Simple past is used to describe an action
-
direction⏎B) equal in magnitude but have opposite direction⏎C) not equal in magnitude but have the same direction⏎D) not equal in magnitude but have opposite direction⏎E) None of the above⏎In a uniform circular motion, which
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
- L15:968 excites
- L15:474 excites
- L15:1081 excites
- L15:532 suppresses
- L12:123 weakly suppresses
- L15:121 weakly excites
- L14:1904 weakly excites
- L15:2915 weakly excites
- L14:580 weakly suppresses
- L7:56 weakly suppresses
- L15:1547 weakly excites "also" before a verb
- L15:1884 weakly excites
- L15:92 weakly suppresses
- L12:689 weakly excites
- L7:2589 weakly excites
- L15:99 weakly excites
- L14:2617 weakly excites
- L14:3069 weakly suppresses
- L6:1258 weakly suppresses "mean by" in questions and definitions
- L15:459 weakly excites
- L6:1781 weakly excites
- L10:3061 weakly excites
- L7:167 weakly suppresses
- L0:450 weakly suppresses
Feeds later layers
- L17:1497 strongly excites
- L17:2570 excites
- L17:1709 excites
- L17:1149 excites
- L18:812 excites
- L17:2419 excites
- L19:1530 excites
- L18:1218 excites
- L17:1586 excites
- L17:1606 suppresses
- L19:742 suppresses
- L18:421 suppresses
- L17:1009 suppresses
- L18:1573 excites
- L19:320 suppresses
- L20:2843 excites
- L19:458 suppresses
- L18:351 suppresses
- L18:981 excites
- L17:1826 excites
- L17:731 excites
- L22:219 suppresses
- L19:200 excites
- L18:1272 excites
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
- лан
- :',⏎
- pecially
- ')),⏎
- ATURE
- ')</
- )'),⏎
- itunes
- ropolitan
- ')):⏎
Suppresses
- !
- !⏎⏎
- !
- !⏎
- !⏎⏎
- ?
- irth
- ?⏎⏎
- 0
- ضبط