Neuron Explorer

Qwen3-0.6B

pronoun 'it' pointing back to a thing

A neuron in layer 14 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 pronouns like 'it', 'they' or 'their' when they refer back to a thing or group mentioned earlier in the sentence. Turning it up made the text drift into unrelated lines, so the output tells us little about this 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

    antzen seems to have been a deacon. The members of this Lithuanian congregation were mostly of Dutch descent; they had also won some Lutherans for their church, which was taken very ill by the Prussian King. The congregation belonged

  2. #2 peak

    It states:⏎All persons born or naturalized in the United States, and subject to the jurisdiction thereof, are citizens of the United States and of the State wherein they reside.Once again the words of the Constitution identify two ways and only two ways that

  3. #3 peak

    Instead of measuring impact by land area like other studies do, theirs uses a new framework to measure the environmental impact of each farming type against the quantity of food that it produces. This revealed that by yield, high-intensity farming was less impactful than

  4. #4 peak

    have come to believe that the whole world is an enigma, a harmless enigma⏎that is made terrible by our own mad attempt to interpret it as though it had⏎an underlying truth."⏎-- Umberto Eco⏎More information about the Python

  5. #5 peak

    see: basically nothing. The town—and “town” is even sort of grand for what it was—has about 8,000 people in it. But here

  6. #6 peak

    it, technology, in the form of an automated house, is portrayed as mindless and soulless, continuing to perform its many functions for the humans who created it but are no longer there.⏎Eventually, the house (technology) is destroyed by

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 think that, maybe it's not the best way to just say that. Maybe you could say, maybe you could say that, maybe that. Maybe, maybe, maybe maybe, maybe, maybe, maybe, maybe, maybe, maybe

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

    flora by means of an appropriate diet, accompanied by a probiotic supplement, offers positive effects in reducing anxiety.⏎Probiotics are microorganisms which are present in their native state in all of us from birth: they’re in the gut, the

  2. #2 peak

    so on the German side, the Enigma cipher system, the Walter propulsion system and of course the German jets, rockets and missiles are just few of the items they came up with. In these pages we'll attempt to explain most technologies relevant to

  3. #3 peak

    design and delivery of policies, programs and services that affect them.⏎How do families living remotely connect?⏎Mornington Island residents have poor quality mobile

  4. #4 peak

    that to U.S. health officials.⏎According to the Centers for Disease Control (CDC), the ALA's N-O-T program is the only smoking cessation program it rates as a "model program." In fact, the CDC does not list any

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. L9:175 suppresses
  2. L8:40 suppresses
  3. L10:1389 suppresses
  4. L4:499 suppresses
  5. L12:2897 excites
  6. L13:1262 suppresses
  7. L12:521 excites
  8. L5:583 excites
  9. L7:2482 suppresses
  10. L4:1306 excites
  11. L8:2143 suppresses
  12. L13:2601 excites
  13. L5:2733 suppresses
  14. L7:2537 suppresses
  15. L13:1997 excites
  16. L7:1905 excites
  17. L10:2920 excites
  18. L6:1922 excites
  19. L11:2714 weakly excites
  20. L1:1871 weakly suppresses
  21. L13:2186 weakly excites
  22. L10:2853 weakly excites
  23. L13:1538 weakly excites
  24. L11:1245 weakly suppresses

Feeds later layers

  1. L16:2272 excites
  2. L15:396 excites
  3. L17:578 excites
  4. L17:1167 suppresses
  5. L17:638 excites
  6. L17:321 excites
  7. L15:714 excites
  8. L19:582 suppresses
  9. L17:314 excites
  10. L17:1070 excites
  11. L15:1226 suppresses
  12. L15:1131 excites
  13. L18:358 weakly excites
  14. L15:1047 weakly suppresses
  15. L15:26 weakly excites
  16. L18:1397 weakly suppresses
  17. L16:1702 weakly excites
  18. L16:577 weakly excites
  19. L15:1127 weakly excites
  20. L17:45 weakly excites
  21. L19:840 weakly excites
  22. L15:2880 weakly suppresses
  23. L19:2261 weakly suppresses
  24. L16:499 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 -3.36 and -3.181 values between -2.10 and -1.921 values between -1.57 and -1.392 values between -1.39 and -1.213 values between -1.21 and -1.033 values between -1.03 and -0.8510 values between -0.85 and -0.6718 values between -0.67 and -0.4937 values between -0.49 and -0.31337 values between -0.31 and -0.132532 values between -0.13 and 0.05755 values between 0.05 and 0.2390 values between 0.23 and 0.4033 values between 0.40 and 0.5824 values between 0.58 and 0.7615 values between 0.76 and 0.9412 values between 0.94 and 1.1210 values between 1.12 and 1.308 values between 1.30 and 1.482 values between 1.48 and 1.662 values between 1.66 and 1.841 values between 1.84 and 2.023 values between 2.02 and 2.192 values between 2.19 and 2.372 values between 2.37 and 2.551 values between 3.09 and 3.271 values between 3.45 and 3.632 values between 3.63 and 3.81offnegativepositiveoften

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. ($)
  2. overrun
  3. KP
  4. overturn
  5. zf
  6. �
  7. __;⏎
  8. ISCO
  9. ($_
  10. Poz

Suppresses

  1. им
  2. him
  3. eren
  4. ODO
  5. iness
  6. ytt
  7. uyện
  8. igkeit
  9. 情况进行
  10. ías