Neuron Explorer

Qwen3-0.6B

"and" joining two actions

A neuron in layer 6 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 "and" when it links two actions or verbs, such as "the markets rallied, the press swooned, and" or "succumb and die." Turning it up did not produce any clear theme in the replies.

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

    Union directives. It is for technical evaluation only. This device has not been authorized as required by the rules of the Federal Communications Commission. This device is not, and may not be offered for sale or lease, or sold or leased until authorization is

  2. #2 peak

    point higher than it is now. Yet, after Nixon’s announcement, the markets rallied, the press swooned, and, even though his speech pre-empted the popular Western Bonanza, the

  3. #3 peak

    it to Willamette’s 1929 freshman orientation. A month later the stock market collapsed — his bank failed, his savings account was wiped out and his scholarship was withdrawn for lack of funds. “I thought the world was coming

  4. #4 peak

    leads to another classic hallmark of Alzheimer's, neurofibrillary tangle in nerve cells. After several decades of exposure to these insults, neurons ultimately succumb and die, leading to progressively damaged learning and memory centers in the brain.⏎The mice

  5. #5 peak

    wartime, when a nation's survival is at stake, the reasons for secrecy are most apparent, the secrecy restrictions imposed by the government are most widespread,[*] and acceptance of those restrictions by the citizens is broadest.[†] In peac

  6. #6 peak

    labor conflict that existed in the past two decades. The AFL and the CIO merged and formed one organization which represented “35% of all nonagricultural workers

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 need to a more comprehensive understanding of the role of the *first* and *second* parts of the problem. In particular, the first part of the problem, I suppose, the need to a more comprehensive understanding of the role of

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

    Tejel (Syria’s Kurds: History, politics and society).⏎Geographically, Kurds in Syria live mostly in non-contiguous regions of the country—as is apparent on the map. Around 30-3

  2. #2 peak

    probably because of water column mixing induced by hydraulic dredging. Overall, 185 phytoplankton taxa were found, of which ca 50% were present in both lakes. No significant differences in species

  3. #3 peak

    with a trained eye.⏎For example, in the image shown below, the bottle’s cast shadow is clearly incongruous with the shape of the bottle (as is the shadow on this cover of Time Magazine). Such obvious errors in a shadow are

  4. #4 peak

    ure, supplemented with need-based fertiliser applications. The most significant aspect of SRI is that the fields are not kept submerged under water all the time, as is usual in rice farming, but are allowed to remain just wet without flooding.⏎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. L1:1696 weakly suppresses
  2. L4:762 weakly suppresses
  3. L4:261 weakly excites
  4. L4:822 weakly suppresses
  5. L4:670 weakly suppresses
  6. L0:3035 weakly excites
  7. L0:1873 weakly excites
  8. L0:1729 weakly excites
  9. L3:569 weakly excites unclear: mid-phrase function words
  10. L2:370 weakly suppresses
  11. L2:2739 weakly excites
  12. L4:438 weakly excites
  13. L3:814 weakly excites
  14. L4:631 weakly suppresses
  15. L1:1121 weakly excites
  16. L0:356 weakly suppresses
  17. L0:618 weakly suppresses
  18. L2:2999 weakly suppresses
  19. L5:460 weakly suppresses
  20. L3:415 weakly excites
  21. L4:486 weakly excites
  22. L3:2674 weakly suppresses
  23. L3:1924 weakly suppresses
  24. L4:2355 weakly excites

Feeds later layers

  1. L8:132 excites
  2. L9:1903 suppresses
  3. L27:717 excites
  4. L7:93 excites
  5. L18:857 excites
  6. L19:1982 excites
  7. L23:298 excites
  8. L12:434 excites
  9. L21:1622 weakly excites
  10. L10:1449 weakly excites
  11. L13:625 weakly excites
  12. L18:509 weakly excites
  13. L20:837 weakly excites
  14. L18:867 weakly excites
  15. L21:655 weakly excites
  16. L18:1519 weakly excites
  17. L8:615 weakly excites
  18. L7:428 weakly suppresses
  19. L17:960 weakly excites
  20. L17:28 weakly excites
  21. L19:1501 weakly suppresses
  22. L18:920 weakly excites
  23. L17:2213 weakly excites
  24. L9:272 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 -2.00 and -1.892 values between -1.07 and -0.962 values between -0.96 and -0.844 values between -0.84 and -0.739 values between -0.73 and -0.6111 values between -0.61 and -0.4925 values between -0.49 and -0.3893 values between -0.38 and -0.26253 values between -0.26 and -0.15814 values between -0.15 and -0.031662 values between -0.03 and 0.09668 values between 0.09 and 0.20225 values between 0.20 and 0.3257 values between 0.32 and 0.4317 values between 0.43 and 0.559 values between 0.55 and 0.6714 values between 0.67 and 0.784 values between 0.78 and 0.906 values between 0.90 and 1.015 values between 1.01 and 1.136 values between 1.13 and 1.256 values between 1.25 and 1.361 values between 1.36 and 1.484 values between 1.48 and 1.594 values between 1.59 and 1.711 values between 1.71 and 1.832 values between 1.83 and 1.941 values between 2.17 and 2.292 values between 2.52 and 2.64offnegativepositiveoften

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. they
  2. others
  3. oster
  4. they
  5. AssignableFrom
  6. 公开发
  7. convo
  8. other
  9. demás
  10. There

Suppresses

  1. affen
  2. 료
  3. ect
  4. ",-
  5. istingu
  6. foresee
  7. /watch
  8. isseur
  9. roducing
  10. reib