Neuron Explorer

Qwen3-0.6B

comma after a sentence subject

A neuron in layer 17 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

The neuron fires on a comma that comes right after the subject of a sentence, such as a person or group name, just before an extra phrase like 'Rama, along with his brother'. Its negative side is a small 'by' or 'for' in phrases like 'by far'. The steering outputs did not show a clear theme.

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

    Hebrides that Columba turned his face when he accepted the terrible penance of Molaise. He bade farewell to his relatives, and, with a few monks who insisted on accompany him whithersoever he might

  2. #2 peak

    built except the roof," Thomas replied. So Gundaphorus sent him gold and silver to roof the palace, and Thomas, thanking<|endoftext|>The Internet and copyright⏎Choose a topic:⏎Internet technology is developing faster

  3. #3 peak

    killed the ten-headed demon Ravana, in Satyug. Ravan had abducted Rama’s wife Sita. Rama, along with his brother Lakshmana, follower Hanuman and an army of monkeys

  4. #4 peak

    Biblical law, Herod did not want to offend the masses (or the Temple priests) by placing his portrait on the kingdom’s coins. Instead, Herod, like his Hasmonean predecessors, decorated his coins with depictions of common

  5. #5 peak

    world.⏎The GIUK Gap forms the principal choke point between Russia’s great Northern Fleet and its strategic interests in the North Atlantic and all points south. The Russians, as a resurgent power, have modernized their military forces, but they still

  6. #6 peak

    om-Tzion had recently died and her sons, Hyrcanus II and Aristobulus II, divided against each other in a terrible civil war.⏎The Romans, as usual, were in the midst of fighting a war. In 64

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: that, that, that, 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

    at low wind strengths:⏎The DualWingGenerator is self-optimising and can adapt itself to different wind conditions. In terms of its efficiency, it is by no means inferior to conventional small wind power stations and features amazing benefits

  2. #2 peak

    ,190 km2 (459 sq mi) with a basal circumference of 140 km. This makes it by far the largest of the three active volcanoes in Italy, being about two and

  3. #3 peak

    rit for Sickness reliever), Azadirachta indica (Neem), belonging to the Meliaceae family, from prehistoric Indian subcontinent, for its multitudinous medicinal properties (Tanguturi et al. 2

  4. #4 peak

    and set the paragraph to be double-spaced. Do not simply hit the Enter button twice at the end of each line, it will become an editing nightmare and when the sorting is finished, it will put all of the blank spaces at the top

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. L10:470 excites
  2. L16:1223 excites
  3. L15:2971 suppresses
  4. L16:1442 suppresses
  5. L7:1098 excites
  6. L15:2835 excites
  7. L14:2402 suppresses
  8. L11:237 suppresses
  9. L13:2366 suppresses
  10. L13:921 excites
  11. L16:1115 suppresses
  12. L16:1621 excites
  13. L15:3052 suppresses
  14. L14:728 suppresses
  15. L14:2416 suppresses
  16. L3:232 excites 'on' after a noun like 'impact on'
  17. L15:804 suppresses
  18. L13:1232 excites
  19. L15:824 suppresses
  20. L7:890 excites
  21. L15:280 excites
  22. L16:1556 excites
  23. L10:1855 suppresses
  24. L5:2747 weakly excites

Feeds later layers

  1. L19:2707 excites
  2. L20:458 excites
  3. L18:1401 suppresses
  4. L24:148 excites
  5. L25:1596 excites
  6. L25:620 excites
  7. L19:2403 excites comma after a full clause
  8. L20:2020 excites comma before a participle phrase
  9. L24:140 excites
  10. L19:3037 weakly excites
  11. L26:252 weakly excites
  12. L21:2114 weakly excites comma before an added phrase
  13. L22:1425 weakly excites
  14. L26:686 weakly excites
  15. L20:349 weakly excites
  16. L19:1995 weakly suppresses comma after a subject phrase
  17. L19:2498 weakly excites
  18. L27:1237 weakly excites
  19. L19:1463 weakly excites
  20. L20:586 weakly suppresses
  21. L18:843 weakly suppresses
  22. L19:2535 weakly excites
  23. L18:1881 weakly excites
  24. L19:818 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.

1 values between -5.58 and -5.232 values between -4.52 and -4.172 values between -3.82 and -3.472 values between -3.47 and -3.122 values between -3.12 and -2.776 values between -2.77 and -2.415 values between -2.41 and -2.0610 values between -2.06 and -1.7114 values between -1.71 and -1.3632 values between -1.36 and -1.0147 values between -1.01 and -0.66164 values between -0.66 and -0.311549 values between -0.31 and 0.051642 values between 0.05 and 0.40319 values between 0.40 and 0.7548 values between 0.75 and 1.1022 values between 1.10 and 1.4510 values between 1.45 and 1.812 values between 1.81 and 2.167 values between 2.16 and 2.516 values between 2.51 and 2.863 values between 2.86 and 3.213 values between 3.21 and 3.563 values between 3.56 and 3.912 values between 3.91 and 4.271 values between 4.27 and 4.621 values between 5.67 and 6.021 values between 6.38 and 6.732 values between 8.13 and 8.48offnegativepositiveoften

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. reflecting
  2. minating
  3. ''' ⏎
  4. 簃
  5. ''' ⏎
  6. ') ⏎
  7. apon
  8. ując
  9. ifr
  10. erot

Suppresses

  1. 控
  2. сел
  3. ц
  4. 腾
  5. lical
  6. 访
  7. stand
  8. 引
  9. 插
  10. colors