the digit 1 in years like 1500s
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. 25 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 digit '1' at the start of a year in the 1400s or 1500s, such as 1533 or 1548. The opposite side reacts to a '2' in years like 2021. Turning it up pushed the model toward number words like 'six' and 'seven', which fits the theme of numbers.
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.
-
(95% CI, 1.4-3.0) for alcohol, 4.2 (95% CI, 3.1-5.8) for marijuana, and
-
Mary Tudor (1516-1558), Mary Stuart (1542-1587), and Elizabeth Tudor (1533-1603).⏎Knox used “monstrous
-
River May (now St. John’s, Florida), in 1564. […]⏎Jacques le Moyne de Morgues (c. 1533–1588) was a French artist and member of
-
profile, head concealed⏎Return to Photo Key to Species in Group B-2b<|endoftext|>Portrait of Katherine Parr, Queen of England (1512-1548)⏎English School, Late Sixteenth Century⏎“As the famous rhyme
-
apples, and sweets in shoes left beside beds, on windowsills, or before the hearth.⏎Colonists came to America after the Reformation in the 1500s. They were primarily Puritans and other Protestant reformers
-
of ours — and the galaxies beyond. And we have Copernicus, among others, to thank for that.⏎Copernicus was born in Poland in 1473 and as a young man became fascinated by math, medicine and science
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: six six seven six seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven seven
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
appearance of troops at Pine Ridge, he advocated for peace with the European Americans⏎Date de dernière mise à jour : 05/07/2021⏎Ajouter un commentaire<|endoftext|>By DANIEL SUDDEATH⏎NEW
-
supports other economic activity, provides an essential product and is poised for long-term growth⏎By Rob Roach, ATB Economics 22 February 202
-
stomach-churning of remedies.⏎A study published on Thursday in The New England Journal of Medicine has found transplanting human faeces into people with the superbug Clostridium difficile provided an almost universal cure for the condition - far outperform
-
ID: 4233) has a total length of 125,982 bp, and it is located in the 7q31 locus of chromosome 7. MET is transcribed into a 6,
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
- L21:1540 excites
- L21:1470 suppresses
- L23:1599 suppresses
- L23:399 excites
- L24:677 suppresses
- L23:1178 excites
- L21:2037 excites
- L22:1016 suppresses
- L22:452 excites
- L24:800 excites colon in clock times
- L24:909 suppresses
- L24:1949 excites
- L23:2848 excites
- L21:1211 suppresses
- L24:2493 excites
- L21:1011 excites
- L24:1627 suppresses
- L24:2295 suppresses
- L22:610 suppresses
- L24:1686 suppresses
- L22:1879 excites
- L23:554 suppresses
- L24:2113 suppresses
- L22:2325 suppresses
Feeds later layers
- L26:898 suppresses
- L26:580 suppresses
- L26:760 suppresses
- L26:880 suppresses
- L26:334 excites
- L27:496 suppresses
- L27:1100 suppresses
- L27:435 suppresses
- L27:205 excites
- L27:494 excites
- L27:785 weakly excites
- L27:678 weakly suppresses
- L26:327 weakly excites
- L26:522 weakly suppresses
- L26:739 weakly suppresses
- L26:1812 weakly suppresses
- L26:905 weakly suppresses
- L27:1448 weakly excites
- L27:391 weakly suppresses
- L26:957 weakly suppresses
- L27:132 weakly excites
- L27:1387 weakly excites
- L27:1232 weakly suppresses
- L27:1409 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
- 线上线下
- 问他
- ícul
- reelection
- 给她
- afen
- 尽快
- 恂
- icion
- .au
Suppresses
- Six
- sixth
- six
- fifth
- arch
- Fifth
- Six
- ��
- Five
- �