Neuron Explorer

Qwen3-0.6B

place and journal names in citations

A neuron in layer 10 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 city, state, or journal name in reference lists, like the comma after 'Portsmouth' in 'Portsmouth, NH' or 'Clinical Chemistry'. It is tuned to bibliography style text. Turning it up made the output drift into unrelated topics, so the steering does not show this clearly.

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

    , Vol 3118, Springer-Verlag, Berlin⏎Heidelberg New York⏎D., Lenoir, R., and McKenzie

  2. #2 peak

    really learn.⏎Beers, Kylene and Robert E. Probst. Notice & Note: Strategies for Close Reading. Portsmouth, NH: Henemann, 2013.⏎Using the work of Louise

  3. #3 peak

    th ed.). Boston, MA: Cengage Learning.⏎[TradingCoachUK]. (2018

  4. #4 peak

    : How to teach writing like you mean it (2nd ed.). Newark, DE: International Reading Association.⏎- Writing activities that help students integrate precise, vivid

  5. #5 peak

    . The Art of Slow Reading. Portsmouth, NH: Heinamann, 2012.⏎And advocate to taking

  6. #6 peak

    azren (2007) Assisting learning using computers through the application of animated pedagogical agents (APA) or intelligent agents. Smart Teaching & Learning: Re-engineering ID, Utilization and Innovation of Technology, 1 .

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: the problem of a single sentence. So, the problem of a sentence. Now, I think about the sentence. Now, I think about the problem of a sentence. So, the problem of the sentence. So, the problem. So

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

    0 percent are either overweight or obese, according to the U.S. Centers for Disease Control and Prevention.⏎Results of the study were published recently in the journal Clinical Chemistry.⏎The U.S. National Heart, Lung, and Blood Institute offers a guide

  2. #2 peak

    Karl M. Johnson, Douglas F. Antczak, William H. Dietz, David H. Martin and Thomas E. Walton appears in Emerging Infectious Diseases 17 (5) (20

  3. #3 peak

    the gastrointestinal system. Their results are published in "Nature Structural & Molecular Biology".⏎"These findings could in principle be explored to develop a decoy alternative to conventional

  4. #4 peak

    for Atmospheric Research and gathered information such as radial velocity and reflectivity from six Doppler weather radars that were in place during storms. Using the Weather Research and Forecasting Model, they found that incorporating the Doppler radar-based information decreased the error

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. L8:2698 suppresses the word 'dental' in oral health
  2. L8:270 suppresses
  3. L7:320 suppresses
  4. L1:1464 weakly excites
  5. L6:55 weakly suppresses
  6. L6:2402 weakly excites
  7. L5:1051 weakly suppresses
  8. L8:1365 weakly suppresses
  9. L8:2792 weakly excites
  10. L5:2321 weakly excites
  11. L4:994 weakly excites
  12. L6:1696 weakly excites
  13. L2:2835 weakly excites
  14. L4:983 weakly suppresses
  15. L1:1319 weakly excites 'for' right after 'look'
  16. L4:1438 weakly excites
  17. L6:2943 weakly suppresses
  18. L1:2801 weakly excites
  19. L2:2147 weakly suppresses
  20. L1:73 weakly suppresses
  21. L1:3003 weakly excites
  22. L1:171 weakly excites "but" after a negated clause
  23. L2:75 weakly suppresses
  24. L7:1163 weakly excites

Feeds later layers

  1. L24:612 weakly excites
  2. L11:2198 weakly suppresses
  3. L12:2948 weakly excites
  4. L12:2082 weakly excites
  5. L15:269 weakly excites
  6. L24:2806 weakly excites
  7. L21:893 weakly suppresses
  8. L11:1762 weakly excites
  9. L12:1580 weakly excites
  10. L11:1720 weakly suppresses
  11. L20:1953 weakly excites
  12. L11:2309 weakly suppresses
  13. L17:1748 weakly suppresses
  14. L11:122 weakly excites
  15. L23:2160 weakly excites
  16. L18:884 weakly suppresses
  17. L14:1185 weakly excites
  18. L16:1684 weakly excites
  19. L16:2978 weakly excites
  20. L15:1764 weakly excites
  21. L25:1650 weakly excites
  22. L14:387 weakly suppresses
  23. L14:842 weakly excites
  24. L17:1844 weakly 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 -1.41 and -1.351 values between -0.91 and -0.851 values between -0.85 and -0.783 values between -0.66 and -0.5910 values between -0.59 and -0.5320 values between -0.53 and -0.4742 values between -0.47 and -0.4183 values between -0.41 and -0.34123 values between -0.34 and -0.28215 values between -0.28 and -0.22336 values between -0.22 and -0.15523 values between -0.15 and -0.09645 values between -0.09 and -0.03707 values between -0.03 and 0.04508 values between 0.04 and 0.10340 values between 0.10 and 0.16160 values between 0.16 and 0.2288 values between 0.22 and 0.2949 values between 0.29 and 0.3526 values between 0.35 and 0.4115 values between 0.41 and 0.486 values between 0.48 and 0.544 values between 0.54 and 0.601 values between 0.60 and 0.661 values between 1.04 and 1.10offnegativepositiveoften

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. cimal
  2. Neill
  3. TokenType
  4. ไท
  5. 缩水
  6. productList
  7. pakistan
  8. Feeling
  9. แยก
  10. فئ

Suppresses

  1. уж
  2. 断
  3. hers
  4. ici
  5. اغ
  6. research
  7. orsch
  8. 塑造
  9. 层面
  10. 让她