the end-of-text token after a finished passage
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. 19 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 invisible marker the training data uses to show that one document has ended and a different one begins. Turning it up gave text that did not obviously relate to this job, so its role beyond the marker is unclear.
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.
-
to ask you another question. What is something you can do well?⏎I’d like to ask you another question. What is something you can’t do well?⏎<|endoftext|>Discover the cosmos! Each day a different image or photograph of our fascinating universe is
-
highlights the extraordinary visual adaptations of miniature marine creatures.⏎CONTENT PROVIDED BY⏎Congdon, O. (n.d.). How is a mantis shrimp’s vision like<|endoftext|>Part 2 - Those Who Are Unable to See the Fact of Creation⏎The
-
. Commonly referred to as the “love hormone,” oxytocin plays a major role in forming bonds and trust with others. Produced in abundance by breastfeeding mothers<|endoftext|>Every week in the art area of the classroom, your toddler will have the opportunity
-
a blood vessel is also called a thrombus. A thrombus can stop blood flow to parts of the body. Thrombosis can cause the following:<|endoftext|>Cry, the beloved country contrasts the rural and urban ways of life. How
-
called a hen. Juvenile turkeys are called poults. A group of turkeys is called a flock.⏎The benefits of raising white turkeys are<|endoftext|>In this thoroughly researched history, Graydon Meints tells the fascinating story of the
-
35 PM EDT. …⏎What is the Flower Moon 2020?⏎The Next Full Moon is the<|endoftext|>“It’s tricky to rock a rhyme⏎to rock a rhyme that’s right on
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 first time I was going to be a 't' in the title of the First People (A in the First People Forward' by the New England, the first time I is to be a 't' or a 'i'
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Geriatric Medicine study are available here: http://www.europeangeriaticmedicine.com/article/S1878-7649(13)00123-X/).⏎It is possible nursing staff underestimate the importance
-
of diabetes and cardiovascular diseases, and conducting continuous surveillance to monitor the trends in NCD prevalence, their risk factors, and treatments.⏎Copyright © 2020 The Royal Society for Public Health. Published by Elsevier Ltd. All rights reserved
-
that the impact of religious education on Taiwanese students' civic values was low, with the main focus being personal and spiritual life development. Copyright © 2021 selection and editorial matter, Kerry J. Kennedy and John Chi-Kin Lee;
-
by dredging activities, along with certain physico-chemical parameters of the water, were the major factors shaping phytoplankton communities.⏎- Abstract views: 1777⏎- PDF: 842⏎Copyright (c
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
- L16:0 suppresses
- L18:2024 excites
- L18:569 excites
- L2:1083 weakly excites
- L3:56 weakly excites
- L0:2552 weakly excites
- L4:618 weakly suppresses
- L12:1178 weakly suppresses
- L3:744 weakly suppresses
- L18:2192 weakly excites
- L15:2363 weakly suppresses
- L6:68 weakly suppresses
- L11:29 weakly excites
- L15:1680 weakly suppresses
- L1:2223 weakly suppresses
- L3:1402 weakly suppresses
- L17:890 weakly excites
- L11:1995 weakly excites
- L8:86 weakly excites
- L0:2297 weakly excites
- L4:426 weakly excites
- L18:2450 weakly excites "all" after a comma in a trailing phrase
- L14:304 weakly suppresses
- L3:1434 weakly suppresses
Feeds later layers
- L24:788 excites
- L22:1685 excites
- L20:1706 weakly excites
- L22:734 weakly excites
- L27:962 weakly excites
- L24:357 weakly excites
- L24:1531 weakly suppresses
- L27:2142 weakly suppresses
- L22:1249 weakly suppresses
- L26:1552 weakly excites
- L26:96 weakly excites
- L20:1921 weakly suppresses
- L25:1532 weakly suppresses 'New' in place names
- L21:217 weakly excites
- L24:1155 weakly excites
- L25:1376 weakly suppresses
- L26:191 weakly suppresses
- L23:89 weakly excites
- L20:1672 weakly suppresses
- L20:2705 weakly suppresses
- L21:1967 weakly suppresses
- L20:1261 weakly excites
- L23:1014 weakly suppresses
- L20:2327 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
- enci
- вен
- uations
- inate
- ém
- xmlns
- ences
- sen
- cbc
- UND
Suppresses
- ipes
- confirm
- akah
- (#)
- ولة
- consistency
- zag
- irror
- confirmation
- ubbles