Neuron Explorer

Qwen3-0.6B

"other" in "on the other hand"

A neuron in layer 22 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 reacts to the word "other" in the phrase "on the other hand", which sets up a contrast. When it was turned up, the model produced long, wandering sentences about unrelated topics, and when turned down it fell into "and and and" loops.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    Convention were drawn up simultaneously, French and English, both being recognized on an equal footing as working languages. The 1929 Conventions, on the other hand, had been drawn up in French only, as French was still the leading

  2. #2 peak

    IM Lander will integrate data from both the TRN system and LIDAR technology to achieve an ultraprecise lunar landing.⏎Navigation GUI, on the other hand, assists in human-machine interactions, making it possible for ground operators to make

  3. #3 peak

    because he banged on about the poor: theologians agree that the Gospels show the influence of Greek tragedy in their structure and tone. Theatre historians, on the other hand, will tell you that the breakthrough came in 1837 with

  4. #4 peak

    word meaning simply “box.” This ark, on the other hand, is represented by a Hebrew word meaning “chest” or “coff

  5. #5 peak

    object-oriented editing the furthest, as the program allows you to open up a clip and access a huge number of clip-related parameters. Sonar, on the other hand, was not designed with object-oriented editing in mind, and takes a bit

  6. #6 peak

    of addressing the overall system in a product’s (or building’s) life, ‘green’ describes single, separate aspects. Sustainable, on the other hand, relates to the whole system encompassed by design, manufacture, and purchase

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: and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and and

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

    greater dilemma posed by<|endoftext|>How can we help?⏎You can also find more resources in our⏎Select a category⏎Something is confusing⏎Something is broken⏎I have a suggestion⏎What is your email?⏎What is 1 +

  2. #2 peak

    ⏎WE BELIEVE that one of the most important subjects of the knowledge of God, is the MONOTHEISM, i.e. the belief that there is but one God. As a matter of fact Monotheism (TOWH

  3. #3 peak

    -reliance, and a highly centralized economic system caused Albania to lag far behind its neighbors in terms of economic development.⏎After Hoxha died in 1985, his hand-picked successor, Ramiz Alia, who became

  4. #4 peak

    , which abruptly brought to an end the global financial system based on the gold standard. The result was that the LMU effectively came to an end in 1914, although it lingered on as a legal entity until its formal dissolution

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. L12:942 excites
  2. L19:2676 weakly excites
  3. L11:1296 weakly suppresses
  4. L3:2972 weakly suppresses word endings in multi-token nouns
  5. L21:226 weakly excites
  6. L12:2878 weakly suppresses
  7. L14:143 weakly excites 'the' inside journal and institution names
  8. L16:473 weakly suppresses
  9. L6:366 weakly suppresses
  10. L5:2442 weakly excites
  11. L20:423 weakly excites
  12. L12:1094 weakly excites
  13. L9:2357 weakly excites
  14. L2:1411 weakly suppresses
  15. L7:469 weakly excites
  16. L14:2394 weakly suppresses unclear: mid-word and descriptive phrases
  17. L10:1452 weakly excites
  18. L1:599 weakly suppresses
  19. L15:508 weakly suppresses
  20. L18:2861 weakly excites
  21. L20:201 weakly suppresses
  22. L0:1351 weakly excites
  23. L1:67 weakly excites
  24. L19:943 weakly excites

Feeds later layers

  1. L24:205 suppresses
  2. L25:6 weakly excites
  3. L25:1598 weakly excites
  4. L25:1740 weakly excites
  5. L26:65 weakly excites
  6. L23:2802 weakly excites
  7. L27:1127 weakly suppresses
  8. L23:958 weakly excites
  9. L27:289 weakly suppresses
  10. L23:507 weakly excites
  11. L27:720 weakly excites
  12. L25:2961 weakly suppresses
  13. L27:2634 weakly suppresses
  14. L23:379 weakly suppresses
  15. L27:556 weakly excites
  16. L24:930 weakly excites
  17. L25:1856 weakly suppresses
  18. L23:1860 weakly excites
  19. L24:1184 weakly suppresses
  20. L23:2120 weakly excites
  21. L27:743 weakly excites
  22. L24:1547 weakly excites
  23. L26:614 weakly excites
  24. L26:1091 weakly excites

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 -11.58 and -10.661 values between -10.66 and -9.743 values between -9.74 and -8.826 values between -8.82 and -7.905 values between -7.90 and -6.9810 values between -6.98 and -6.0618 values between -6.06 and -5.1428 values between -5.14 and -4.2263 values between -4.22 and -3.30104 values between -3.30 and -2.38189 values between -2.38 and -1.46394 values between -1.46 and -0.532106 values between -0.53 and 0.39522 values between 0.39 and 1.31193 values between 1.31 and 2.2394 values between 2.23 and 3.1563 values between 3.15 and 4.0730 values between 4.07 and 4.9926 values between 4.99 and 5.9112 values between 5.91 and 6.8312 values between 6.83 and 7.759 values between 7.75 and 8.679 values between 8.67 and 9.593 values between 9.59 and 10.512 values between 10.51 and 11.432 values between 12.35 and 13.271 values between 13.27 and 14.191 values between 24.31 and 25.23offnegativepositiveoften

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. oshi
  2. ness
  3. sẵn
  4. avail
  5. ARY
  6. ********************************************************************************
  7. 可想
  8. disp
  9. depth
  10. alysis

Suppresses

  1. 和服务
  2. and
  3. RequestId
  4. 和
  5. lation
  6. _-
  7. ству
  8. apsulation
  9. And
  10. jud