article 'a' after 'here are'
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. 3 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 "a" in the phrase "Here are a few..." and similar list introductions, where a list is about to start. Turning it up made the text drift into unrelated, repetitive wording rather than any clear list.
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.
-
ocular depth cues aid the detection of moving objects by moving observers.<|endoftext|>Cut flowers need proper care to prolong the length of time you can enjoy them. Here are a few suggestions:⏎- Pick cut flowers in the morning when stems and leaves are t
-
SLEEP HYGIENE TIPS:⏎- Create a sleep ritual (a special set of little things) to do before bed to relax the body. Here are a few ideas: stretching, meditation, breathing exercises or gratitude journaling.⏎- Create
-
mental health. Here are a few ways giving back can benefit you, just as much as the recipients of your
-
good teachers.⏎The above research is supported by a previous study I reported on in, How to Increase Economic Output by $112 Trillion! Here are a couple quotes from that post:⏎Effect of Good Teachers on Students Lifetime Earnings⏎
-
harm to our health and wellbeing is clear. You can probably already feel these effects and are looking to find ways to cut down on your screen time. Here are a few simple steps to overcoming the problem of screen time and embracing a digital detox.⏎
-
to know what to do to get your middle school child ready for all of the activities, requirements and events that will go on during the school year. Here are a few tips that you can follow that may come in handy as you ready your youngster
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: the number of 2000. The numbers 2000. The numbers are. the numbers are. the numbers are. the numbers are. the numbers are. the numbers are. the numbers are. the numbers are
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
0 with 997 athletes (22 of them women) from 24 countries competing in 95 events. The U
-
devastating fire. Smart cities can also help alleviate traffic and address environmental concerns.⏎The Internet of Things (IoT) is a sensor network made up of billions of different smart devices. From small household appliances to large industrial tools, these objects all have
-
9 major planets around the sun: Mercury, Venus, the Earth, Mars, Jupiter, Saturn, Uranus, Neptune and Pluto. Only the first five of them (not mentioning the Earth) have been known for thousands of years by many civilizations
-
on the U.S. Team in the 1936 Olympic Games held in Berlin, Germany. They won 14 medals — eight of them gold: a quarter of the 56 medals won by the entire U.S
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
- L2:149 strongly suppresses
- L2:1426 excites
- L2:502 excites 'other' after a preposition
- L1:683 excites
- L2:1120 suppresses
- L2:295 weakly excites
- L2:1002 weakly excites
- L2:2177 weakly suppresses
- L0:954 weakly excites
- L2:631 weakly suppresses
- L1:523 weakly excites
- L1:2493 weakly excites
- L0:56 weakly excites
- L2:326 weakly suppresses
- L0:2914 weakly suppresses
- L0:69 weakly suppresses
- L0:314 weakly suppresses
- L2:840 weakly excites
- L2:1390 weakly excites
- L1:2407 weakly suppresses
- L1:1793 weakly excites 'he' in pheasant, 'mouth' in Dartmouth
- L1:1691 weakly suppresses
- L0:632 weakly excites
- L2:1061 weakly suppresses
Feeds later layers
- L6:156 strongly excites
- L4:450 strongly suppresses
- L10:661 strongly excites
- L17:2744 strongly excites 'of' after 'those' or 'these'
- L7:257 strongly excites
- L5:478 strongly excites
- L16:1317 strongly excites
- L7:146 strongly excites
- L10:246 strongly excites
- L10:1190 strongly suppresses
- L18:1096 strongly excites
- L14:1008 strongly excites
- L13:535 strongly excites
- L9:2445 strongly excites
- L11:1684 strongly excites
- L4:298 strongly excites
- L19:1629 strongly suppresses
- L13:1675 strongly suppresses
- L6:542 strongly excites
- L16:1418 strongly suppresses
- L11:2210 strongly excites
- L17:1696 strongly excites
- L21:275 strongly suppresses
- L8:688 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
- 各自的
- 各自
- 也都
- elves
- którzy
- paces
- köp
- aren
- èrent
- _are
Suppresses
- itself
- Its
- Its
- its
- vention
- iene
- halfway
- uario
- 率达
- behind