Neuron Explorer

Qwen3-0.6B

the letter r in the middle of words

A neuron in layer 27 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 letter r in the middle of a word, such as the r in Avranches, Citrullinemia, midrashic, paraphrased and lacrimation. When turned up, the model drifted into odd loops like 'everything everything', which does not match the letter pattern.

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

    Norman cathedral. It is the seat of the Bishop of Coutances and Avranches, previously the Bishops of Coutances. Standing 80 metres (

  2. #2 peak

    button to use the lights. A specimen can be tested under multiple lights at the same time. The instrument offers highly accurate and reliable test results<|endoftext|>What is Citrullinemia Type 1?⏎Citrullinemia type I is a

  3. #3 peak

    Mary Potter Engel⏎Midrash refers to the rabbinic tradition of interpretation that explores scripture beyond its literal meanings. The verse by verse interpretations generated by midrashic commentary bind the Talmud,

  4. #4 peak

    "Just a moment," and turned to the subject and repeated the same request. If the subject asked for additional information, the experimenter repeated, or paraphrased the request. If the subject persisted, the experimenter said

  5. #5 peak

    .⏎tAsAM ruditashabdena gavAM haMbhAraveNa cha |⏎vrajasyotthApanaM chakrurghoShavR^iddhAH samAgatAH ||2

  6. #6 peak

    Llangynhafal, Llanychan, Llanbedr D.C. and Llanrhydd. Llannerch contained the parishes of Llanfair D

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:icicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicicic

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

    The phase-difference can, therefore, be defined as ΔΦ=ΦV-ΦI.⏎We define the instantaneous power similarly to the DC power by P(t)=V(t)×I(t). When using

  2. #2 peak

    power frequency in the United States), calculate impedances for all components and determine all voltage drops using Ohm's Law (E=IZ ; I=E/Z ; Z=E/I). The polar magnitudes of the results should closely agree with your vol

  3. #3 peak

    for all the areas.⏎To address this problem, we just need to recreate categories for all the features. First, delete the single category from them.⏎v.category input=gadm36_d

  4. #4 peak

    ="500"> <polygon points="200,10 250,190 160,210" style="fill:lime;stroke:yellow;stroke-width:1"/> </svg

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. L21:3031 excites
  2. L25:459 suppresses
  3. L20:777 suppresses
  4. L25:252 excites
  5. L23:2074 suppresses
  6. L19:89 excites
  7. L2:202 excites "of" after "many" or "which"
  8. L0:117 suppresses
  9. L23:233 suppresses
  10. L23:310 excites
  11. L22:1753 suppresses
  12. L20:895 excites
  13. L22:1306 excites
  14. L8:3 excites
  15. L8:7 excites
  16. L20:747 excites
  17. L18:569 excites
  18. L24:938 suppresses
  19. L26:606 suppresses
  20. L26:1564 excites
  21. L16:41 suppresses
  22. L25:956 excites
  23. L8:16 weakly suppresses
  24. L8:14 weakly excites

Feeds later layers

None: this is the last layer.

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.

6 values between -3.66 and -1.423898 values between -1.42 and 0.831 values between 0.83 and 3.081 values between 7.57 and 9.821 values between 79.45 and 81.701 values between 83.94 and 86.19offnegativepositiveoften

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. =(
  2. ={{
  3. =ax
  4. =F
  5. =\"
  6. ={
  7. ="
  8. =[],
  9. ]="
  10. =s

Suppresses

  1. ic
  2. icit
  3. am
  4. icos
  5. onne
  6. ons
  7. ame
  8. icot
  9. icol
  10. ов