Neuron Explorer

Qwen3-0.6B

'for' after 'known' or 'renowned'

A neuron in layer 1 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron lights up on the word 'for' when it follows praise words like 'known' or 'renowned', as in 'known for his charity'. Turning it up made the model write loose, rambling stories that did not clearly relate to the prompt.

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 activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

  1. #1 peak

    ensure that the process of arbitration was carried out properly. All of 13th century Christian Europe willingly looked upon him as an international judge.⏎He was renowned for his charity. “The peace and blessings of the realm come to us through the

  2. #2 peak

    Attacks⏎The Dog Breed Known As Great Danes⏎Great Danes are widely assumed to be the biggest of all dog breeds around the globe. They’re renowned for their majesty and imposing presence. While they are often gentle in nature and even

  3. #3 peak

    short history has resulted in a rich blend of cultural and traditional customs. All Americans are renowned for their openness and friendliness to visitors. As a rule, dress is casual;

  4. #4 peak

    of Information in 1918, and especially as Minister of Aircraft Production in 1940. He is best known for his newspaper empire, including the Express and Evening Standard newspapers which gained record circulation.

  5. #5 peak

    , politician and civil rights leader. He was known for striving to better relationships between black residents and the police force, and he was a

  6. #6 peak

    a streetscape panel displayed on Roybal campus showing the original and expanded areas of CDC’s campus.<|endoftext|>Woods Hole Oceanographic Institution (WHOI), renowned for its role in the discovery of the Titanic, continues on its mission of research at

Steering

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.

turned down a lot
Prompt

Here is something I have been thinking about lately: I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

  1. #1 peak

    man of the people.” It’s even odder that he’s accused of being a “leftist.” In fact, he’s the son of a wealthy rancher once accused of killing leftist leaders, whose bodies were found hidden on the family ranch

  2. #2 peak

    2 1 min read⏎It’s always a good day to thank farmers, ranchers and agri-food workers because they literally keep us alive, but it makes extra

  3. #3 peak

    merer Early Horse⏎Utah Museum of Natural History⏎Courtesy & © Mary Heers, PhotographerIn 1928 Elmer Cook, a rancher in Hagerman, Idaho, noticed an interesting bone sticking our of the hill

  4. #4 peak

    permits, those permits would be subject to challenge in federal court, delaying or halting these activities resulting in a huge impact on rural economies.⏎Farmers and ranchers do a good job taking care of the land. As I often say, they

Connections

Show on map

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Fed by earlier layers

  1. L0:1301 suppresses
  2. L0:178 suppresses
  3. L0:1224 weakly suppresses
  4. L0:2676 weakly suppresses
  5. L0:488 weakly excites
  6. L0:679 weakly suppresses
  7. L0:494 weakly suppresses
  8. L0:572 weakly excites
  9. L0:1411 weakly excites
  10. L0:21 weakly suppresses last digit of years like 1970
  11. L0:3009 weakly excites
  12. L0:2347 weakly excites
  13. L0:1960 weakly excites unclear: mixed medical and harsh words
  14. L0:1391 weakly excites
  15. L0:2606 weakly excites
  16. L0:193 weakly excites
  17. L0:950 weakly excites
  18. L0:1356 weakly suppresses
  19. L0:1555 weakly suppresses
  20. L0:2930 weakly excites
  21. L0:2128 weakly excites
  22. L0:2440 weakly excites
  23. L0:482 weakly excites
  24. L0:1722 weakly excites

Feeds later layers

  1. L10:3001 weakly suppresses
  2. L7:369 weakly suppresses
  3. L15:300 weakly excites
  4. L2:435 weakly suppresses
  5. L18:3025 weakly suppresses
  6. L11:2249 weakly suppresses
  7. L20:857 weakly suppresses
  8. L2:2062 weakly suppresses
  9. L13:1936 weakly excites
  10. L7:1567 weakly suppresses
  11. L11:2462 weakly suppresses
  12. L27:1580 weakly suppresses
  13. L6:536 weakly excites
  14. L3:1136 weakly suppresses
  15. L9:2700 weakly excites
  16. L19:2678 weakly excites
  17. L13:2198 weakly suppresses
  18. L23:2275 weakly suppresses
  19. L12:1603 weakly excites
  20. L21:2159 weakly suppresses
  21. L17:2817 weakly suppresses
  22. L8:2435 weakly suppresses
  23. L4:1151 weakly suppresses
  24. L20:2180 weakly suppresses

Activation histogram

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.

2 values between -1.07 and -0.991 values between -0.99 and -0.911 values between -0.91 and -0.831 values between -0.83 and -0.753 values between -0.75 and -0.672 values between -0.67 and -0.597 values between -0.59 and -0.5110 values between -0.51 and -0.4312 values between -0.43 and -0.3520 values between -0.35 and -0.2766 values between -0.27 and -0.19255 values between -0.19 and -0.11921 values between -0.11 and -0.031519 values between -0.03 and 0.05725 values between 0.05 and 0.13251 values between 0.13 and 0.2170 values between 0.21 and 0.2914 values between 0.29 and 0.377 values between 0.37 and 0.455 values between 0.45 and 0.536 values between 0.53 and 0.611 values between 0.69 and 0.771 values between 0.77 and 0.851 values between 0.93 and 1.011 values between 1.09 and 1.171 values between 1.17 and 1.251 values between 1.32 and 1.401 values between 1.72 and 1.803 values between 2.04 and 2.12offnegativepositiveoften

Logit lens

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 layers instead.

Promotes

  1. coli
  2. hood
  3. Walters
  4. 举
  5. avit
  6. V
  7. ogy
  8. ppv
  9. Charg
  10. feather

Suppresses

  1. ermann
  2. Nederland
  3. elper
  4. reinc
  5. useParams
  6. 男朋友
  7. .invalid
  8. 自然是
  9. \application
  10. หลวง