'only' in 'the only'
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. 4 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 reacts to the word 'only' when it comes right after 'the' or 'weren't the', as in 'the only person'. Turning it up made the model's writing repeat the word 'only' and similar phrasing, though the effect is loose.
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.
-
food diet is true to its name – most, if not all, foods need to be “raw”. This means dieters are prohibited from cooking food. The only allowed “cooking” is
-
of Valois, although Henry hired a professional poet to do the actual writing.⏎By the 1700s, kings and dukes weren’t the only ones dispensing valentines. Each February 14, handwritten ‘Val
-
connecting the satellite system to the backbone of land-based internet.⏎Reddit user andrew867 noted in his post that "Eastlink has direct access to the only viable dark fibre/wavelength service coming into the province from Nova Scotia. Zayo
-
the Bronze Age, to ancient Alexandria and the founding of the Silver Throne - in one night. And in that single night, John and Jack will find that the only way to save their friend and stop the chaos destroying the world is to solve a
-
Park?⏎The only person I know who has ever seen a snake in Central Park is the writer Roger
-
been a peculiar battle. Both sides feel strongly on the matter, but even in the most crucial moments, there has been little uncontained anger. Indeed, the only thing that has been expended at all recklessly in the long fight has been talk
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: for example, if I have a 100, and the number of elements is 10, then I have $ 10 \times 10 $. 1. 10 10 10
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
her to the fence. Her head was held high, and she was blowing. I kept her there until she relaxed. I never did find out what was the matter. Maybe a deer? This is just one example of a horse spooking
-
. Gates then used a computer at the University of Washington to produce the traffic flow charts. (Paul Allen's father was a librarian at UW.) This was the beginning of Traf-O-Data.⏎The next step was to build a device to
-
they frequently used sunscreen, however, had 23% greater risk of multiple sunburns in the past year compared to people who said they seldom used the stuff.⏎Those who frequently sought shade and pulled on long sleeves, however, had about
-
| Volume 25, Issue 1⏎That was the end of the campaigning in 1776. Washington went into winter quarters and
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
- L3:2168 excites
- L3:344 excites
- L3:122 weakly excites
- L2:205 weakly excites
- L2:1390 weakly suppresses
- L3:59 weakly excites
- L2:330 weakly excites
- L3:115 weakly suppresses
- L1:1674 weakly excites
- L3:800 weakly suppresses
- L3:2287 weakly excites
- L1:6 weakly excites
- L3:2890 weakly suppresses
- L2:3065 weakly excites
- L0:1734 weakly excites
- L3:2664 weakly excites
- L2:2715 weakly suppresses
- L0:71 weakly excites
- L1:1280 weakly suppresses
- L2:243 weakly suppresses
- L2:2149 weakly suppresses
- L2:2152 weakly suppresses
- L0:1640 weakly excites
- L2:2891 weakly excites
Feeds later layers
- L5:613 excites
- L11:2781 excites
- L10:568 excites
- L20:2485 excites
- L12:637 excites
- L14:2791 weakly excites
- L11:1216 weakly excites
- L7:2432 weakly excites
- L6:1209 weakly excites
- L5:1440 weakly excites
- L17:2967 weakly excites
- L5:1532 weakly excites
- L17:1975 weakly excites
- L6:2985 weakly excites
- L18:1951 weakly excites
- L16:1128 weakly excites
- L15:839 weakly suppresses
- L15:2568 weakly excites
- L7:2660 weakly excites
- L21:2442 weakly excites
- L8:432 weakly suppresses
- L15:2849 weakly excites
- L17:2069 weakly excites
- L6:625 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
- 最关键
- atl
- 最关键的
- 次要
- 都不是
- acon
- 最重要的
- 来找
- lé
- instance
Suppresses
- 族自治州
- нер
- velope
- rogram
- muş
- 热情
- clus
- alyzer
- utenberg
- ):⏎⏎