Neuron Explorer

Qwen3-0.6B

the piece 'Mont' in place names

A neuron in layer 24 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 start 'Mont' of place names such as Montenegro, Montpellier and Montevideo. Turning it up made the model repeat 'a little a little', which has no clear link to the place names.

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

    Egypt, El Salvador, Finland, Grenada, Guatemala, Japan, Lao, PDR, Macedonia, former Yugoslav Republic of, Mongolia, Montenegro, New Zealand, Nicaragua, Nigeria, Norway, Pakistan, Paraguay,

  2. #2 peak

    16. A year later, the French settled in the area just north of present Montogmery and built Fort Toulouse on the river.⏎The Coushatta

  3. #3 peak

    istinguished Service Cross, the Croix de Guerre with palms, the French Legion of Honor, the Croce di Guerra of Italy, the War Medal of Montenegro, and ultimately, the Congressional Medal of Honor in March of 19

  4. #4 peak

    of yellowfin tuna (Thunnus albacares) in nearshore areas in French Polynesia, determined using ultrasonic telemetry⏎Centre IRD de Montpellier, BP 5045, 34032 Mont

  5. #5 peak

    Republic’s School of Agronomy through an agreement with the government of the province of Montevideo and the Primary Education Board.⏎The programme provides the opportunity for direct contact

  6. #6 peak

    was born in Montpelier, Vermont, on December 26, 1837

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 right and the left in the ricki ricki ricki ricki ricki ricki ricki ricki ricki ricki ricki ricki ricki rick

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

    natural enemies that help suppress pest densities."⏎The study, "Multi-State Trials of Bt Sweet Corn Varieties for Control of the Corn Earworm (Lepidoptera: Noctuidae)," analyzed the performance of Bt sweet corn

  2. #2 peak

    of events. Fate defines events as ordered or "inevitable" and unavoidable. Classical and European mythology features three goddesses dispensing fate, known as Moirai in Greek mythology, as Parcae in Roman mythology, and as

  3. #3 peak

    prominent Creole family, the Puente de Ayala was opened to traffic in 1880. It connected the San Miguel district to Malate and Ermita.⏎The bridge also served Isla de Convalescencia, the

  4. #4 peak

    leshana tova (happy New Year), to send greeting cards, and perhaps plant trees in Israel in their honor.⏎Although the high holidays lack the merriment of Purim, the gift giving of Hanukkah, and the joy

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. L20:435 excites
  2. L21:1720 excites
  3. L20:2520 weakly excites
  4. L18:1779 weakly suppresses
  5. L21:1379 weakly excites
  6. L19:1057 weakly excites
  7. L21:1555 weakly suppresses
  8. L18:1056 weakly suppresses
  9. L19:788 weakly suppresses
  10. L6:482 weakly suppresses
  11. L2:2055 weakly suppresses
  12. L0:655 weakly suppresses
  13. L7:3057 weakly excites
  14. L15:878 weakly suppresses
  15. L18:2042 weakly excites
  16. L1:1251 weakly suppresses
  17. L18:406 weakly excites
  18. L19:2972 weakly suppresses
  19. L18:1731 weakly excites
  20. L1:152 weakly excites
  21. L6:2547 weakly excites
  22. L16:2855 weakly suppresses
  23. L15:1308 weakly suppresses
  24. L1:1882 weakly suppresses

Feeds later layers

  1. L25:1586 excites
  2. L27:663 excites
  3. L27:797 excites
  4. L27:231 suppresses
  5. L25:45 suppresses
  6. L26:484 excites
  7. L26:411 suppresses
  8. L27:709 excites
  9. L25:1132 weakly suppresses
  10. L25:2182 weakly excites
  11. L26:213 weakly suppresses
  12. L27:854 weakly excites
  13. L26:1102 weakly suppresses fibromyalgia and hour counts
  14. L26:918 weakly excites
  15. L26:911 weakly suppresses
  16. L27:343 weakly suppresses
  17. L27:383 weakly suppresses
  18. L26:2923 weakly excites
  19. L27:2093 weakly excites
  20. L25:1048 weakly excites
  21. L25:2748 weakly suppresses
  22. L27:173 weakly suppresses
  23. L26:2293 weakly excites
  24. L27:1637 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.

4 values between -12.34 and -11.711 values between -10.43 and -9.791 values between -9.79 and -9.151 values between -9.15 and -8.511 values between -8.51 and -7.873 values between -7.87 and -7.232 values between -7.23 and -6.605 values between -6.60 and -5.964 values between -5.96 and -5.321 values between -5.32 and -4.684 values between -4.68 and -4.045 values between -4.04 and -3.4014 values between -3.40 and -2.769 values between -2.76 and -2.1324 values between -2.13 and -1.4975 values between -1.49 and -0.85690 values between -0.85 and -0.212603 values between -0.21 and 0.43358 values between 0.43 and 1.0755 values between 1.07 and 1.7114 values between 1.71 and 2.355 values between 2.35 and 2.986 values between 2.98 and 3.626 values between 3.62 and 4.264 values between 4.26 and 4.901 values between 4.90 and 5.541 values between 5.54 and 6.184 values between 6.18 and 6.823 values between 7.46 and 8.091 values between 8.09 and 8.731 values between 9.37 and 10.011 values between 10.01 and 10.651 values between 12.56 and 13.20offnegativepositiveoften

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. a
  2. 一个
  3. asmine
  4. 一个小
  5. 一个问题
  6. ือน
  7. .makeText
  8. FFECT
  9. uality
  10. 是一個

Suppresses

  1. erto
  2. ientos
  3. BTS
  4. elier
  5. Gb
  6. URNS
  7. concentr
  8. Dana
  9. nämlich
  10. taire