Neuron Explorer

Qwen3-0.6B

'for' in research institute names

A neuron in layer 3 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 'for' inside the names of institutes and organisations, such as 'Institute for Research' or 'Institute for Animal Health'. Turning it up did not produce anything obviously related, so the steering test tells us little about it.

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

    variety of reasons, including the study of human genetic diseases,'' said Dr. Gregory M. Acland, a research scientist at the James A. Baker Institute for Animal Health at

  2. #2 peak

    the plan doesn’t answer is how the city will cope with the climate change that has already been set in motion.⏎Enter a study prepared by Oakland’s Pacific Institute for the California Energy Commission, published in July but not widely circulated until this month.

  3. #3 peak

    Other agents have also been moved to the highest risk category following the report.⏎The findings that led to the decision by the World Health Organization (WHO) International Agency for Research on Cancer (IARC) to reclassify the cancer risk of sunbed

  4. #4 peak

    and trusted source of scientific evidence of what works in education. The WWC is administered by the Department, through a contract to a joint venture of the American Institutes for Research and the Campbell Collaboration.⏎To provide information needed by decision-makers, the W

  5. #5 peak

    four times more likely than older drivers to crash for every mile driven.⏎The graduated licensing programs are meant to prevent this.⏎James Fell and colleagues at the Pacific Institute for Research and Evaluation in Maryland examined data on fatal crashes from 1990

  6. #6 peak

    Linzhao Cheng, Ph.D., an associate professor of gynecology and obstetrics and co-director of the stem cell program of the Johns Hopkins Institute for Cell Engineering. “This is one area of stem cell biology where human and mouse

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: if I have a problem of a book, a book, a book, and I have a problem, a problem, a problem, and a problem. I have a problem of a book, a book, a book, and I have

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

    , It is too quiet on the streets below.⏎2. I wish we were not stuck in the house all the time, said Winnie.⏎3. Wilbur did not like his special collar.⏎4. The scientist on the TV said you

  2. #2 peak

    "⏎- Brix Orolfo, age 12yrs.⏎- © Brix Orolfo⏎

  3. #3 peak

    Ocean. Ukrainian Cossacks formed the Zaporozhia Cossack State in 1649. They are considered

  4. #4 peak

    office for women at the State Department, to Amina Benkhadra, Morocco’s Minister of Energy, Mines, Water and Environment, to Christiana Figueres, executive secretary of the U.N. framework on Climate Change, women are

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. L0:1543 excites
  2. L2:1622 weakly excites
  3. L1:1341 weakly suppresses
  4. L1:2581 weakly excites
  5. L1:1258 weakly excites
  6. L0:1474 weakly excites
  7. L0:340 weakly suppresses
  8. L1:1179 weakly excites 'in' after 'result' (results in)
  9. L0:1939 weakly suppresses
  10. L1:1710 weakly excites
  11. L2:2422 weakly suppresses
  12. L2:2351 weakly suppresses
  13. L1:813 weakly suppresses
  14. L0:2044 weakly excites
  15. L1:529 weakly excites
  16. L2:1539 weakly suppresses
  17. L1:1793 weakly suppresses 'he' in pheasant, 'mouth' in Dartmouth
  18. L0:644 weakly suppresses
  19. L2:500 weakly suppresses
  20. L1:2740 weakly suppresses word fragments inside non-English words
  21. L2:2558 weakly suppresses
  22. L0:694 weakly suppresses 'seen' after 'never' or 'rarely'
  23. L2:2656 weakly excites
  24. L2:1321 weakly suppresses

Feeds later layers

  1. L19:1221 strongly excites
  2. L5:1090 excites
  3. L17:1223 excites
  4. L16:2336 excites
  5. L16:524 excites
  6. L9:2587 excites
  7. L10:1029 excites
  8. L7:1172 excites 'for' after past participles like selected
  9. L4:1035 excites 'for' after verbs like 'fall' or 'click'
  10. L4:1653 excites
  11. L18:1028 excites
  12. L16:463 excites 'for' introducing a purpose or target
  13. L17:2949 weakly excites 'for' introducing an example or reason
  14. L16:2690 weakly suppresses
  15. L23:2014 weakly excites
  16. L19:1237 weakly excites
  17. L23:2386 weakly excites
  18. L18:1818 weakly suppresses quantifier 'all' mid-phrase
  19. L9:1890 weakly excites
  20. L21:474 weakly suppresses
  21. L7:2026 weakly excites
  22. L18:2783 weakly excites
  23. L6:1880 weakly suppresses "by" introducing a deadline or means
  24. L15:2925 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.

3 values between -1.55 and -1.321 values between -1.32 and -1.082 values between -0.85 and -0.6257 values between -0.62 and -0.391035 values between -0.39 and -0.152552 values between -0.15 and 0.08207 values between 0.08 and 0.319 values between 0.31 and 0.553 values between 0.55 and 0.782 values between 0.78 and 1.012 values between 1.01 and 1.241 values between 1.24 and 1.483 values between 1.48 and 1.711 values between 1.94 and 2.171 values between 2.64 and 2.871 values between 3.34 and 3.575 values between 3.57 and 3.801 values between 3.80 and 4.041 values between 4.04 and 4.271 values between 4.27 and 4.501 values between 4.73 and 4.972 values between 4.97 and 5.201 values between 5.20 and 5.431 values between 5.43 and 5.664 values between 5.66 and 5.903 values between 5.90 and 6.131 values between 6.13 and 6.362 values between 6.36 and 6.602 values between 6.60 and 6.831 values between 7.06 and 7.292 values between 7.53 and 7.76offnegativepositiveoften

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. geries
  2. lops
  3. purposes
  4. bidden
  5. 目的
  6. amen
  7. yı
  8. tran
  9. andes
  10. pastoral

Suppresses

  1. egot
  2. ENSIONS
  3. eña
  4. ᴇ
  5. �
  6. ess
  7. 意义上
  8. e
  9. ENSION
  10. 以习近平