Neuron Explorer

Qwen3-0.6B

'of' after 'learned' or 'because'

A neuron in layer 8 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 fires on the word "of" right after verbs or links like "learned of" and "because of". Turning it up did not produce any clear theme in the model's writing, so the link to this word is only seen in the examples.

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

    helped, but I still felt weak, low in energy and mentally fuzzy.”⏎Then her husband, Richard Brown, a psychopharmacologist at Columbia University, learned of a Russian herb, Rhodiola rosea, which purportedly strengthened the whole

  2. #2 peak

    Ocean Science (NCCOS) explain.⏎“We did this study because of concerns that putting marine finfish farms in the coastal ocean could have adverse effects on

  3. #3 peak

    ative” (signed “story”) is sometimes preferred over “oral” because of an enduring history of “oral” education that has powerful, often painful, con

  4. #4 peak

    It was early in the history of the Seneca when they first came across the spot where a black, sticky substance was rising from the water. They quickly learned of its potential as body paint, medicinal purposes, and as additive for making the flare

  5. #5 peak

    Belgium⏎The rare disease affects the lungs, causing worries about the future for Fernanda.⏎Fernanda Aspilche can remember the exact day she learned of her diagnosis with Alpha 1 Antitrypsin Deficiency, a

  6. #6 peak

    in your mouth and chew the health benefits of almonds with joy.⏎Getting healthier by understanding nutrition is one of my main goals. I love eating almonds, especially because of how good they are for you. Research more benefits of almonds to satisfy all of

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: what is the value of the given question is the of the given question is to determine whether this is a question of the question asked, or is it that the question is of being the first of identifying the the the question is of being the

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

    , a task often complicated by poor-to-nonexistent Cold War recordkeeping;⏎- To prepare a waste management and cleanup plan that identifies

  2. #2 peak

    shape of things to come for women will be shorter and plumper, and it's all thanks to natural selection

  3. #3 peak

    northerners, but two of the invading peoples settled on the coast of Palestine, between Gaza and Mount Carmel. The attempted invasion ended Egyptian pretensions to a Syro-Palestinian hegemony.⏎Two more years of peace ensued, but

  4. #4 peak

    so that she is certain not to go out of the lines while another little girl with the same princess coloring pages scribbles quickly in a hurry to get done first.⏎

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. L6:1133 strongly suppresses "to" before an infinitive verb
  2. L6:1880 strongly suppresses "by" introducing a deadline or means
  3. L6:2747 strongly excites
  4. L7:681 strongly suppresses 'from' after prevent or stop
  5. L6:439 excites
  6. L6:914 suppresses
  7. L6:2064 excites
  8. L6:2359 suppresses
  9. L5:1755 suppresses
  10. L6:502 excites
  11. L4:113 excites
  12. L6:2020 suppresses
  13. L6:1621 excites
  14. L6:2313 suppresses
  15. L6:2000 excites
  16. L7:111 suppresses
  17. L6:2882 excites
  18. L6:2057 excites
  19. L4:2908 suppresses
  20. L3:1128 excites
  21. L4:518 excites
  22. L4:698 suppresses
  23. L3:1787 excites
  24. L3:307 excites

Feeds later layers

  1. L19:385 strongly excites
  2. L16:1153 suppresses
  3. L9:226 suppresses 'to' starting an infinitive
  4. L18:715 suppresses
  5. L18:472 suppresses
  6. L18:452 suppresses
  7. L15:1176 suppresses
  8. L18:141 excites
  9. L14:659 suppresses
  10. L22:200 suppresses
  11. L19:1509 suppresses
  12. L17:1592 suppresses
  13. L19:617 suppresses
  14. L26:1122 suppresses
  15. L19:652 suppresses
  16. L17:1994 suppresses
  17. L18:1027 suppresses
  18. L21:299 suppresses
  19. L17:1196 excites
  20. L16:2726 suppresses
  21. L24:208 suppresses
  22. L13:909 suppresses 'to' before a verb after a noun
  23. L12:1000 suppresses
  24. L11:429 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.

1 values between -1.65 and -1.531 values between -1.41 and -1.294 values between -1.29 and -1.182 values between -1.18 and -1.061 values between -1.06 and -0.946 values between -0.94 and -0.8211 values between -0.82 and -0.7034 values between -0.70 and -0.5860 values between -0.58 and -0.46166 values between -0.46 and -0.34316 values between -0.34 and -0.22542 values between -0.22 and -0.101512 values between -0.10 and 0.02795 values between 0.02 and 0.14226 values between 0.14 and 0.2691 values between 0.26 and 0.3843 values between 0.38 and 0.4930 values between 0.49 and 0.6125 values between 0.61 and 0.739 values between 0.73 and 0.852 values between 0.85 and 0.9713 values between 0.97 and 1.094 values between 1.09 and 1.215 values between 1.21 and 1.332 values between 1.33 and 1.451 values between 1.45 and 1.572 values between 1.57 and 1.691 values between 1.69 and 1.811 values between 2.04 and 2.161 values between 2.16 and 2.281 values between 3.00 and 3.12offnegativepositiveoften

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. ように
  2. 理想
  3. role
  4. 旁
  5. 素养
  6. aside
  7. те
  8. 潜力
  9. 理想的
  10. lương

Suppresses

  1. APTER
  2. 回事
  3. 把它
  4. amentos
  5. 办
  6. ῶ
  7. quences
  8. zion
  9. ophon
  10. thereof