Neuron Explorer

Qwen3-0.6B

colon after a metadata label

A neuron in layer 2 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 lights up on the colon after labels in catalogue-style records, such as "Published:", "Date:" or "Publication Date:". Turning it up did not make the writing look like those records.

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

    Science Grants by Hughes Institute⏎By WILLIAM CELIS 3d⏎Published: October 4, 1994⏎An $86 million grant

  2. #2 peak

    801-1869⏎Articles of Agreement and Specifications, Topeka School House Construction⏎Authors: New England Emigrant Aid Company⏎Date: June 9, 1857⏎Charles H. Branscomb

  3. #3 peak

    0.3390/ijerph9114170⏎Published: 15 November 2012⏎Abstract: Understanding the effects of

  4. #4 peak

    Cook, James (Capt.)⏎Subject: Kerguelen Islands, Christmas Island⏎Period: 1785-87 (circa)⏎Publication: Troisieme Voyage De Cook, ou Voyage a l'Ocean Pacifique…⏎

  5. #5 peak

    results⏎Authors: Walker, Robert J. (John), 1801-1869⏎Date: May 27, 1857⏎In this long and formal

  6. #6 peak

    an attachment loop.⏎Date: Circa 1st-3rd century AD Condition: Extremely fine. The

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 is the name of the object and the value of the expression or the result statement and the combination of the result statement and the m and the n and the c and the grouping statement and the m and the n and the c and the

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

    ⏎But here thing is, save() method can return that primary key id value which is generated by hibernate and we can see it by⏎long s = session.save(k);⏎In this same case, persist() will never give any value

  2. #2 peak

    10 0 20. I hope i have been clear enough. ------------------ Come on in !! Drinks are on the house in the Big Moose Saloon !!⏎Udayan Naik<BR>Sun Certified Programmer for the Java

  3. #3 peak

    around the world once more to witness another Venus Transit. It's the last that any of us will see

  4. #4 peak

    system, and multithreading.⏎Also included is coverage of some of Java's most powerful features,⏎such as generics, lambda expressions, and the very important⏎modules feature. A number of recent additions to the

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. L0:1559 excites
  2. L0:1981 weakly suppresses
  3. L0:185 weakly excites
  4. L0:9 weakly suppresses
  5. L1:2944 weakly suppresses
  6. L0:2632 weakly suppresses
  7. L1:1340 weakly suppresses
  8. L1:2638 weakly excites
  9. L1:465 weakly suppresses
  10. L1:2714 weakly excites
  11. L1:1777 weakly excites
  12. L0:334 weakly suppresses
  13. L1:1393 weakly excites
  14. L0:2685 weakly excites
  15. L0:1197 weakly suppresses
  16. L1:1887 weakly excites
  17. L1:2674 weakly suppresses
  18. L1:1620 weakly suppresses
  19. L1:1877 weakly suppresses
  20. L0:1591 weakly suppresses
  21. L1:1918 weakly suppresses
  22. L1:190 weakly excites
  23. L1:1543 weakly excites
  24. L1:535 weakly suppresses

Feeds later layers

  1. L17:343 suppresses
  2. L6:378 suppresses
  3. L20:431 suppresses
  4. L20:936 suppresses
  5. L26:1014 suppresses
  6. L4:368 suppresses
  7. L21:448 suppresses
  8. L9:195 suppresses
  9. L25:1298 suppresses
  10. L4:2962 suppresses
  11. L24:1550 suppresses
  12. L23:917 suppresses the digit '1' starting a year
  13. L25:1461 suppresses
  14. L4:1388 excites
  15. L24:1850 suppresses
  16. L3:88 suppresses
  17. L24:479 weakly suppresses
  18. L22:567 weakly suppresses
  19. L19:989 weakly suppresses
  20. L17:388 weakly excites
  21. L22:2975 weakly suppresses
  22. L21:2052 weakly suppresses
  23. L4:310 weakly suppresses
  24. L7:433 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 -0.93 and -0.871 values between -0.50 and -0.448 values between -0.44 and -0.389 values between -0.38 and -0.3229 values between -0.32 and -0.26112 values between -0.26 and -0.20265 values between -0.20 and -0.14530 values between -0.14 and -0.08953 values between -0.08 and -0.011058 values between -0.01 and 0.05587 values between 0.05 and 0.11234 values between 0.11 and 0.1780 values between 0.17 and 0.2321 values between 0.23 and 0.297 values between 0.29 and 0.353 values between 0.35 and 0.413 values between 0.41 and 0.471 values between 0.47 and 0.531 values between 0.53 and 0.601 values between 0.66 and 0.721 values between 0.72 and 0.781 values between 0.78 and 0.841 values between 1.39 and 1.451 values between 1.45 and 1.51offnegativepositiveoften

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. 的一员
  2. 当晚
  3. 受益
  4. 对她
  5. .Debugger
  6. 렴
  7. 的问题
  8. cmath
  9. CID
  10. 請求

Suppresses

  1. uncate
  2. osi
  3. sealing
  4. orges
  5. otlin
  6. @Override
  7. riding
  8. andscape
  9. urga
  10. нести