Neuron Explorer

Qwen3-0.6B

spot just before a year or century

A neuron in layer 26 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 spot in a sentence where a year or century number is about to come, such as after 'in the' before '18th century' or 'in' before '2008'. When turned up or down, the model's written output did not show a clear matching theme.

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

    at the Uffizi and finally in Palazzo Pitti with the foundation of the Accademia del Cimento in 1657. In the 18th century the Lorraine family kept collecting and the Grand Duke Peter Le

  2. #2 peak

    replace them.⏎- This low-fertility regime resulted from the fall in the total fertility rate that followed the postwar baby-boom (1946 to 19

  3. #3 peak

    Believe it or not, the sauce was brought to France by an Italian chef called Catherina de Medici in the 16th century and was originally served as a veloute (with lots of

  4. #4 peak

    atory to the Benelux Customs Union along with the Netherlands and Luxembourg. This later became the Benelux Economic Union in 1958 and in 2008 the treaty between the three countries was renewed and revised under the

  5. #5 peak

    area. The great tower or keep is the oldest part of the castle and retains many interesting features.⏎Later remodelling, especially in the 19th century, has added many decorative elements to the castle although retaining much

  6. #6 peak

    ), total government spending at all levels was about 8% of gross domestic product. By 1940, it was approximately 15%. In 2010, it had risen to 36%.

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: what if our world could generate energy both green and green energy? Green energy both green energy and green energy both green energy both green energy both green energy both green energy both green energy both green energy both green energy both green energy both green energy

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

    said Lopez at the site, where the temperature reaches 45 C (113 F) at this time of year.⏎The first investigatory work was

  2. #2 peak

    to the space<|endoftext|>SAFETY & ESCAPE PLAN⏎BEFORE A VIOLENT INCIDENT OCCURS⏎• Teach your children to call 911.⏎• Identify neighbors you can tell about the violence in your life and ask

  3. #3 peak

    the more secure and complex WPA2, which became available in 2004 and is a common shorthand for the full IEEE 802.11i (or IEEE 802.11i-20

  4. #4 peak

    15.00 hours, chances are you wouldn't immediately recall. But if you were asked where you were when you heard the news about 9/11 you would probably instantly recall. This is because you have repeated this experience many

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. L25:492 strongly excites
  2. L25:2960 strongly suppresses
  3. L22:217 strongly suppresses
  4. L25:1452 strongly excites
  5. L25:1478 strongly suppresses
  6. L23:617 strongly excites
  7. L23:1598 strongly excites
  8. L25:845 strongly suppresses
  9. L22:844 strongly suppresses
  10. L23:416 strongly suppresses
  11. L24:2078 strongly excites
  12. L24:979 strongly excites
  13. L21:1882 strongly suppresses
  14. L22:2405 strongly excites
  15. L19:1985 strongly excites
  16. L22:1564 strongly excites
  17. L24:413 strongly suppresses
  18. L22:1577 strongly suppresses
  19. L21:1670 strongly suppresses
  20. L21:626 strongly suppresses
  21. L21:2039 excites
  22. L25:2806 suppresses
  23. L24:2759 suppresses
  24. L25:2442 excites

Feeds later layers

  1. L27:180 strongly excites
  2. L27:328 strongly suppresses
  3. L27:902 strongly suppresses
  4. L27:294 strongly excites
  5. L27:132 suppresses
  6. L27:1522 suppresses
  7. L27:678 excites
  8. L27:530 excites
  9. L27:992 excites
  10. L27:1100 excites
  11. L27:205 suppresses
  12. L27:1153 weakly excites
  13. L27:862 weakly excites
  14. L27:893 weakly excites
  15. L27:1157 weakly suppresses
  16. L27:1422 weakly suppresses
  17. L27:1087 weakly excites
  18. L27:448 weakly excites
  19. L27:1329 weakly excites
  20. L27:153 weakly suppresses the spot just before a historical number
  21. L27:352 weakly excites
  22. L27:435 weakly excites
  23. L27:572 weakly excites
  24. L27:1202 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 -23.25 and -21.252 values between -17.26 and -15.262 values between -13.26 and -11.261 values between -11.26 and -9.273 values between -9.27 and -7.277 values between -7.27 and -5.2715 values between -5.27 and -3.2723 values between -3.27 and -1.282173 values between -1.28 and 0.721672 values between 0.72 and 2.723 values between 2.72 and 4.721 values between 8.71 and 10.711 values between 10.71 and 12.712 values between 12.71 and 14.711 values between 32.68 and 34.681 values between 54.66 and 56.66offnegativepositiveoften

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. يث
  2. AccessException
  3. 耻
  4. ReferentialAction
  5. embar
  6. ecome
  7. adamente
  8. bble
  9. 局限
  10. ثق

Suppresses

  1. 1
  2. ₁
  3. 1
  4. �
  5. mười
  6. ¹
  7. Eins
  8. errorHandler
  9. 🔟
  10. 十一