'in' in 'take in' phrases
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. 2 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 fires on the word 'in' when it follows 'take' or 'taking', as in 'take in fluids' or 'taking in the news'. Its negative side is weak and shows up on pieces of longer words. Turning it up did not produce a clear topic in the output.
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.
-
.⏎What’s more, taking in emotional news (whether positive or negative) from social media or playing online games can
-
flow are what keep a business going. What good are record revenues if you’re spending more money than you’re taking in?⏎Sidebar – My buddy Steve Chou owns a successful online wedding linens store
-
body can survive with only water but it isn’t a good idea. Not only is there an absence of taste it also means that your body has nothing to take in. Pregnancy Safe Digestive Enzym
-
may be unable to take in a sufficient amount of food and fluids. It can cause a weight loss of more
-
get started slimming down a healthy diet is with ingesting significantly less food items. Numerous research shows that the level of calories from fat that you take in immediately effects exactly how much unwanted fat retailers you retail store. Grows your levels
-
The Value of Everything: Making and Taking in the Global Economy by Mariana Mazzucato. Copyright © 20
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 need to take this to the end, I can't take it to the end. I have to take it to the end. I can't take it to the end. I can't take it to the end. I can
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
, yes?)⏎It’s super easy to get started too.⏎Here’s a quick how-to:⏎- Get in a comfy position. You don’t need to pretzel it up, know lying down can lead to sleeping, but other than that,
-
- Peter (also called Simon Peter) established the first church in Antioch and is regarded as the founding pope of the Catholic church. Instrumental in
-
, which was located in Florence at that period. He ultimately rose to the rank of papal secretary. Pope Nicholas V employed him as a translator of Greek patristic and pagan authors; his retranslations from Aristotle became particularly influential. Trebiz
-
patristic texts?⏎1. Did Mary merit to be the Mother of God?⏎2
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
- L1:1955 weakly excites
- L1:367 weakly suppresses
- L1:1960 weakly excites
- L1:2459 weakly excites
- L1:658 weakly excites
- L1:2646 weakly excites
- L1:189 weakly suppresses
- L0:1355 weakly excites
- L1:1481 weakly excites
- L0:48 weakly excites
- L1:87 weakly suppresses
- L1:908 weakly excites
- L1:1350 weakly excites
- L1:2922 weakly excites
- L1:514 weakly excites
- L1:1370 weakly excites
- L1:124 weakly suppresses
- L1:2801 weakly excites
- L0:1224 weakly suppresses
- L0:97 weakly excites
- L0:873 weakly suppresses
- L0:25 weakly excites
- L0:1038 weakly excites
- L1:1735 weakly excites
Feeds later layers
- L23:329 suppresses
- L3:2385 weakly excites
- L25:1474 weakly suppresses
- L23:2913 weakly suppresses
- L14:2064 weakly suppresses
- L4:1397 weakly suppresses
- L3:2929 weakly suppresses
- L26:2683 weakly suppresses
- L9:2775 weakly excites
- L9:1371 weakly excites
- L6:2545 weakly suppresses
- L19:890 weakly excites
- L17:2763 weakly suppresses
- L21:608 weakly suppresses
- L20:2539 weakly excites
- L23:1405 weakly suppresses
- L21:1736 weakly excites
- L16:921 weakly excites
- L23:754 weakly suppresses
- L22:2978 weakly suppresses
- L17:1302 weakly suppresses
- L13:2677 weakly excites
- L27:14 weakly suppresses
- L15:1983 weakly 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
- .readline
- ingham
- 尽
- landers
- isible
- ly
- ent
- ASH
- ith
- k
Suppresses
- 台上
- xbc
- 该院
- instituted
- ailer
- instit
- Cubs
- oser
- ulan
- Aunt