Neuron Explorer

Qwen3-0.6B

'only' before a verb or adjective

A neuron in layer 18 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 word 'only' when it sits in the middle of a phrase like 'is only effective' or 'are only just beginning', right after a verb. It is a hedging word before the main point. Turning it up gave lists of counting or repeated descriptive phrases, which does not clearly match the trigger.

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

    cal infection, which causes grave illnesses, including pneumonia, meningitis, and middle-ear infections. Children and the elderly are especially at risk. Vaccines are only effective against a few of the pneumococcal types and increasing resistance to antibiotics

  2. #2 peak

    of blah magazine’, cardboard red and blue ’3D’ glasses?! These classics marked the beginning of mainstream 3D for film and TV audiences – in fact only recently there were a spate of classic 3D movies played on Channel

  3. #3 peak

    the canal became an issue. So in 1905 the city completed the bridge over the canal. The first bridge was a transporter bridge, on of only a few ever built, and had a gondola which was raised and transported

  4. #4 peak

    “Overall, young people’s ability to reason about the information on the Internet can be summed up in one word: bleak.”⏎The question of how it works is only slightly less obvious. The short answer is, “Mud sticks”. If you

  5. #5 peak

    the aptitude and the inclination should have the opportunity to receive entrepreneurial training and access to resources like low-interest loans to become self-employed.⏎Of course, training is only the first step. Once trained, they should be connected job opportunities. Life skill

  6. #6 peak

    its native ecosystems means a return to a more natural state, a point we made in a post last week. Second, ecosystem dynamics are incredibly complex and we are only just beginning to understand the role that even a single species can play.⏎Wolves

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 best of the best (or the best of the best of the best) is therefore (A) A (A) A (A) A (A) A (A) A (A) A (A) A (A

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

    19 percent).⏎Among Gulf War-era II veterans in 2009, 31 percent of veterans aged 18 to 24 years were enrolled in school. As for older

  2. #2 peak

    and 10% with colorectal cancer.⏎Those three cancers accounted for slightly more than half of all cancer diagnoses.⏎Nearly 60% of survivors were 65 or older, the report found. The largest number of survivors with prostate

  3. #3 peak

    in history,” citing the lack of students being exempted from the draft as the reasoning.⏎But nothing is said about the fact that less than half of the population ages 18-26 are actually considered. If the military wanted to have

  4. #4 peak

    rivers, and in small streams, lakes, marshes, and even roadside ditches containing adequate year-round water flow. In areas where deep, calm water is not available, beavers that have enough building material available will create ponds by building dams

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. L15:2956 excites
  2. L16:1867 suppresses
  3. L13:2011 excites
  4. L17:1746 suppresses
  5. L14:1201 excites
  6. L14:1150 suppresses
  7. L16:2328 suppresses
  8. L14:835 suppresses
  9. L15:2592 suppresses
  10. L16:390 suppresses
  11. L14:2639 weakly excites
  12. L15:1963 weakly excites
  13. L14:2689 weakly suppresses
  14. L14:1459 weakly excites
  15. L17:3004 weakly suppresses
  16. L12:2655 weakly excites
  17. L15:826 weakly suppresses
  18. L6:1149 weakly suppresses
  19. L12:2754 weakly suppresses
  20. L8:343 weakly suppresses
  21. L13:1365 weakly excites
  22. L13:2879 weakly suppresses
  23. L14:770 weakly suppresses
  24. L12:1917 weakly excites

Feeds later layers

  1. L19:2188 excites
  2. L19:2926 excites
  3. L19:1179 suppresses
  4. L21:2415 weakly suppresses
  5. L21:1226 weakly suppresses
  6. L24:2068 weakly suppresses
  7. L19:1151 weakly excites
  8. L22:2290 weakly excites
  9. L27:620 weakly excites
  10. L21:956 weakly suppresses
  11. L22:758 weakly excites
  12. L19:1131 weakly excites
  13. L19:3025 weakly suppresses
  14. L19:2909 weakly suppresses
  15. L21:2742 weakly excites
  16. L25:437 weakly suppresses
  17. L25:478 weakly suppresses
  18. L21:1589 weakly excites
  19. L19:1717 weakly suppresses
  20. L19:1019 weakly excites
  21. L25:1813 weakly suppresses
  22. L23:377 weakly suppresses
  23. L20:1923 weakly excites
  24. L19:2006 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 -7.80 and -7.451 values between -7.45 and -7.102 values between -7.10 and -6.751 values between -6.75 and -6.403 values between -6.04 and -5.692 values between -5.69 and -5.341 values between -5.34 and -4.994 values between -4.99 and -4.649 values between -4.64 and -4.2913 values between -4.29 and -3.948 values between -3.94 and -3.5815 values between -3.58 and -3.2319 values between -3.23 and -2.8820 values between -2.88 and -2.5339 values between -2.53 and -2.1852 values between -2.18 and -1.8388 values between -1.83 and -1.48111 values between -1.48 and -1.13174 values between -1.13 and -0.77317 values between -0.77 and -0.42762 values between -0.42 and -0.071424 values between -0.07 and 0.28393 values between 0.28 and 0.63196 values between 0.63 and 0.9877 values between 0.98 and 1.3457 values between 1.34 and 1.6931 values between 1.69 and 2.0428 values between 2.04 and 2.3924 values between 2.39 and 2.749 values between 2.74 and 3.098 values between 3.09 and 3.444 values between 3.44 and 3.799 values between 3.79 and 4.151 values between 4.15 and 4.501 values between 4.50 and 4.852 values between 5.55 and 5.902 values between 5.90 and 6.25offnegativepositiveoften

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. oningen
  2. 缓缓
  3. mons
  4. powers
  5. ineff
  6. 精心
  7. 随着
  8. tiny
  9. getDoctrine
  10. tap

Suppresses

  1. closest
  2. None
  3. Neither
  4. [X
  5. only
  6. Neither
  7. overall
  8. erland
  9. [Y
  10. еть