Neuron Explorer

Qwen3-0.6B

job and work words

A neuron in layer 25 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron tends to fire on words about jobs and work, such as 'job', 'work', and 'workbooks', as well as the word 'City' in a university name. Turning it down made the model keep repeating the word 'class', which this neuron appears to hold back.

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

    1 (1983): 4–43.⏎Wood, Christopher S. Albrecht Altdorfer and the Origins of Landscape. Chicago: University of Chicago Press, 1993.⏎Saint George and the Dragon

  2. #2 peak

    classroom settings, of course. Inattentive mistakes can show up in the workplace, in organizing your day-to-day life – or in particularly inconvenient settings like job interviews.⏎And people can always take those mistakes to be representative of your overall competence

  3. #3 peak

    ayashi is Associate Professor in the Department of Media and Communication at City University of Hong Kong.⏎Atsushi Tago is Professor in the School of

  4. #4 peak

    them. This makes the process of study planning and career development more concrete and measurable.⏎Students can report from their interests, career goals and experiences in internships and work. This helps explore possible career paths and formulate concrete steps to achieve them.⏎SL

  5. #5 peak

    jobs available to African Americans. The plan, which targets communities with persistently high unemployment, includes three main components: creation of public sector jobs, job training with job-placement programs, and wage subsidies for employers. Although the plan is constructed with African

  6. #6 peak

    in all five boroughs of NYC, Yonkers, Westchester, Nassau and Suffolk.⏎CITE PD: CITE offers CTLE-approved in-school or online professional development tailored to your school’s needs and your

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: class class class, class class class, class class class, class class class. Class class class, class class class, class class class, class class class, class class class. Class class class, class class class, class class class,

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

    half a century.⏎On June 15, 1977, former slave Henry Ossian Flipper graduated from West Point with the ranking of second lieutenant in the all-African-American 10th Cavalry.⏎While he

  2. #2 peak

    coat belonged to Charles Johnston Tinkham, Lt. Col. of the 26th Illinois Volunteers. Born in Covington, Kentucky, Tinkham graduated from West Point Academy in 1849, and worked as a civil

  3. #3 peak

    : Institutional⏎American Society of Civil Engineers Award of Merit 2015⏎Civic Trust Award 2014⏎Civic Trust Awards Special Award for Sustainability 2014⏎ENR Best Global Education/Research

  4. #4 peak

    commandant" or "adjutant," and to be composed of two companies, each consisting of a captain, one lieutenant, two ensigns, one first sergeant,

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. L7:92 suppresses
  2. L20:2742 suppresses
  3. L9:207 weakly excites
  4. L3:130 weakly suppresses
  5. L16:1220 weakly suppresses
  6. L4:3038 weakly suppresses
  7. L23:798 weakly excites
  8. L1:25 weakly suppresses
  9. L6:1673 weakly excites
  10. L11:39 weakly excites
  11. L18:66 weakly excites
  12. L8:2252 weakly excites
  13. L11:223 weakly suppresses
  14. L16:141 weakly excites
  15. L16:257 weakly suppresses
  16. L4:1363 weakly excites
  17. L6:1267 weakly excites
  18. L1:2920 weakly suppresses
  19. L4:92 weakly excites
  20. L6:2783 weakly suppresses
  21. L8:84 weakly suppresses
  22. L7:771 weakly excites
  23. L5:948 weakly excites digits inside years like 19xx
  24. L4:342 weakly excites

Feeds later layers

  1. L27:834 suppresses
  2. L26:1517 suppresses
  3. L27:1682 weakly suppresses
  4. L27:285 weakly suppresses
  5. L26:941 weakly suppresses
  6. L26:2745 weakly excites
  7. L26:739 weakly suppresses
  8. L27:2382 weakly excites
  9. L27:2763 weakly excites
  10. L27:1001 weakly excites
  11. L26:2374 weakly suppresses
  12. L27:2295 weakly excites
  13. L26:2427 weakly suppresses
  14. L27:1060 weakly excites
  15. L27:2361 weakly suppresses
  16. L26:2479 weakly suppresses
  17. L26:2593 weakly excites
  18. L26:1055 weakly excites
  19. L26:329 weakly excites
  20. L26:1158 weakly suppresses
  21. L27:2397 weakly excites
  22. L26:2350 weakly excites
  23. L26:1997 weakly excites
  24. L27:923 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 -33.94 and -32.351 values between -19.63 and -18.041 values between -18.04 and -16.451 values between -16.45 and -14.865 values between -14.86 and -13.262 values between -13.26 and -11.677 values between -11.67 and -10.0810 values between -10.08 and -8.4925 values between -8.49 and -6.9029 values between -6.90 and -5.3149 values between -5.31 and -3.72129 values between -3.72 and -2.13421 values between -2.13 and -0.542899 values between -0.54 and 1.05173 values between 1.05 and 2.6473 values between 2.64 and 4.2328 values between 4.23 and 5.8220 values between 5.82 and 7.418 values between 7.41 and 9.0011 values between 9.00 and 10.595 values between 10.59 and 12.182 values between 12.18 and 13.773 values between 13.77 and 15.361 values between 15.36 and 16.951 values between 20.13 and 21.722 values between 21.72 and 23.311 values between 28.08 and 29.67offnegativepositiveoften

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. tridge
  2. 靡
  3. 子
  4. че
  5. maxY
  6. isons
  7. .setMax
  8. max
  9. شن
  10. setMax

Suppresses

  1. class
  2. Class
  3. Class
  4. class
  5. classes
  6. (class
  7. -Class
  8. -class
  9. getClass
  10. getClass