Neuron Explorer

Qwen3-0.6B

semicolons joining two clauses

A neuron in layer 17 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 a semicolon that joins two full sentences into one, usually where the second part explains or results from the first. Its top examples are very consistent, but turning it up or down only produced unrelated text.

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

    that winners of plurality primaries may be weaker nominees than ones who ran in primaries with a majority requirement; and 2) a relatively high number of open seats in districts that are safe

  2. #2 peak

    to the Indian Forest Act, 1927. The amendments have been laid down with the primary objective of abolishing the 1927 Act; as it was a reproduction of the colonial era and a sign of the oppression perpetrated

  3. #3 peak

    method first eliminates any anatomical features not associated with the lung parenchyma; the tissues involved in gas exchange serve as the main sites of COVID-19

  4. #4 peak

    their operation. Also, the hood will have better airflow across its work surface if there are minimal numbers of bottles, beakers, and laboratory apparatus inside the hood; therefore, it is prudent to keep unnecessary equipment and glassware outside of the hood

  5. #5 peak

    early whites and gums healthy.⏎3. Having one or more oral piercings⏎You may think keeping up with latest style trends like piercings is chic; however, it can wreak havoc

  6. #6 peak

    can lead to stroke. When the arteries narrow, this condition is called “narrowing.”⏎Arterial pressure may lead to a weakening of the blood vessels; this is what causes bloodshed inside the brain. Bad habits, such as smoking

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: 2004; 2004-1; 2004-1-1 (first-order) ?; 2004-1-1-1-1... ?; and so on

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

    also exploring a potential new therapy that could target H3K27me3 and reverse the lowered levels. This work is still<|endoftext|>This coin represent King Eukratides I of Bactria, a land in central Asia (Afghan

  2. #2 peak

    prison or a $250,000 fine.⏎And the system has not been updated since women were allowed to serve in combat roles.⏎The Selective Service System has disbanded

  3. #3 peak

    and this is one of the most beautiful coins of the late Hellenistic period. Known for their strongly realistic portraits, this tetradrachm of Eukratides is no exception. He is depicted as a formidable warrior, wearing a

  4. #4 peak

    But except for the state of Texas, women cannot be drafted.⏎Within 30 days of turning 18, all male citizens must register for the Selective Service System,

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. L5:1522 suppresses
  2. L3:452 excites
  3. L2:975 excites
  4. L6:7 weakly suppresses
  5. L16:1040 weakly excites
  6. L3:1163 weakly excites
  7. L3:115 weakly suppresses
  8. L8:7 weakly excites
  9. L16:262 weakly suppresses
  10. L16:28 weakly suppresses
  11. L2:387 weakly excites
  12. L1:331 weakly suppresses
  13. L0:1192 weakly suppresses
  14. L5:75 weakly excites
  15. L1:1262 weakly suppresses
  16. L16:96 weakly suppresses
  17. L4:854 weakly excites
  18. L1:957 weakly suppresses
  19. L0:904 weakly excites
  20. L8:16 weakly suppresses
  21. L16:170 weakly suppresses
  22. L16:145 weakly excites
  23. L4:51 weakly excites
  24. L3:104 weakly excites

Feeds later layers

  1. L18:1465 strongly excites semicolon joining clauses
  2. L20:2064 strongly excites
  3. L21:2311 suppresses
  4. L21:351 suppresses
  5. L18:2832 excites
  6. L27:347 suppresses
  7. L19:1198 suppresses
  8. L24:668 weakly excites
  9. L27:1672 weakly excites
  10. L18:822 weakly suppresses
  11. L26:2886 weakly suppresses
  12. L26:421 weakly suppresses
  13. L24:378 weakly suppresses
  14. L23:171 weakly excites
  15. L22:666 weakly suppresses
  16. L26:416 weakly suppresses
  17. L26:1447 weakly suppresses
  18. L19:1147 weakly suppresses
  19. L27:1042 weakly suppresses
  20. L20:435 weakly suppresses
  21. L20:336 weakly excites
  22. L21:233 weakly excites
  23. L19:1815 weakly suppresses
  24. L19:893 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.

228 values between -2.37 and -1.123647 values between -1.12 and 0.1414 values between 0.14 and 1.394 values between 1.39 and 2.652 values between 2.65 and 3.901 values between 3.90 and 5.162 values between 6.42 and 7.671 values between 13.95 and 15.201 values between 18.97 and 20.231 values between 23.99 and 25.251 values between 26.50 and 27.761 values between 37.80 and 39.061 values between 39.06 and 40.311 values between 40.31 and 41.571 values between 44.08 and 45.332 values between 46.59 and 47.84offnegativepositiveoften

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. however
  2. `}⏎
  3. 本报讯
  4. atically
  5. }/
  6. 年代
  7. angel
  8. //⏎
  9. }//
  10. },⏎

Suppresses

  1. ;!
  2. ;</
  3. ();)
  4. lsen
  5. ;
  6. ;&#
  7. ;p
  8. ;<
  9. ”;
  10. ;'>