Neuron Explorer

Qwen3-0.6B

conjunction before the last list item

A neuron in layer 4 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on the word 'and' (or a similar word like 'also') when it comes right before the final item of a list. Steering did not produce a clear, consistent change in the writing.

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

    and controlling the direction of negotiations; preparing court reports, social case histories, correspondence and other documents; implementing legislative enactments and court rules relating to a case; and keeping abreast of current trends, rules and legislation.⏎Although many mediators are

  2. #2 peak

    its victim, an American tourist who apparently died after being stung by the species while swimming in Australia; a charismatic rhinoceros beetle from the Amazon; and the "Michelin Man" plant from Australia.⏎The Mindoro stripe-faced fruit

  3. #3 peak

    100 hours of supervised driving; a minimum age of 17 for a provisional license; restrictions on driving at night; limits on other teenage passengers; and a minimum age of 18 for a full license.⏎"These findings on nighttime

  4. #4 peak

    interrogative taught as a skill to assist beginning students in language use outside the classroom; using oral histories to promote literacy skills; using the newspaper with ESL students; and preparing the student for employment. A final report of the project is appended to the

  5. #5 peak

    and carefully written text. Not just that it covers important concepts in a well-detailed way, it also tells them quickly with the help of diagrams and bullets. To get the book,

  6. #6 peak

    the flag bearer, while the other soldiers were sculpted from photographs of posed Confederate soldiers. Postponed from May

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 first time, so I need to do this. Wait, so I don't do that. But wait, maybe I'm not sure. Let me also check. Wait, so I don't do that. Okay, but maybe I

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

    direct taxes and indirect taxes ?⏎15. Explain the diminishing rate of substitution.⏎16. Give any two differences between Micro and Macro – economics.⏎17. Draw

  2. #2 peak

    before bed can have a detrimental impact on your sleep?⏎Here, we delve into the negative effects of screen time on the quality of your night’s rest. We also share some tips on how to improve your sleep pattern by reducing phone usage before you

  3. #3 peak

    neutralize certain chemicals that can damage cells. Treating the mice with a compound that mops up these chemicals restores their brain connections, the study shows. It also improves the performance of the mice on cognitive tests.⏎“It not only fixes the number

  4. #4 peak

    harness the immune system’s power to fight prostate, breast and kidney cancers and melanoma.⏎This fact sheet from the National Cancer Institute explains the subject thoroughly. It also lists a number of the clinical trials currently underway on experimental therapeutic vaccines.<|endoftext|>What

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. L3:542 strongly excites
  2. L3:490 strongly excites
  3. L3:403 excites
  4. L3:308 excites
  5. L3:792 suppresses
  6. L3:419 excites
  7. L3:96 excites
  8. L3:234 excites
  9. L3:345 suppresses
  10. L3:564 suppresses
  11. L2:719 suppresses
  12. L3:775 excites
  13. L3:674 excites
  14. L2:182 suppresses
  15. L3:351 suppresses
  16. L2:942 suppresses
  17. L1:107 suppresses
  18. L3:517 excites
  19. L2:936 suppresses
  20. L2:713 suppresses
  21. L3:633 suppresses
  22. L2:227 excites
  23. L3:557 suppresses
  24. L3:274 weakly excites

Feeds later layers

  1. L5:286 excites
  2. L5:43 suppresses
  3. L5:473 excites
  4. L7:796 suppresses
  5. L7:25 suppresses
  6. L7:181 suppresses
  7. L6:47 suppresses
  8. L7:94 excites
  9. L7:43 excites
  10. L7:787 excites
  11. L5:177 excites
  12. L8:1885 suppresses
  13. L15:2967 weakly suppresses
  14. L8:169 weakly suppresses
  15. L5:49 weakly suppresses
  16. L8:2812 weakly excites
  17. L9:1831 weakly suppresses
  18. L6:1152 weakly suppresses
  19. L16:1075 weakly suppresses
  20. L8:142 weakly suppresses
  21. L8:510 weakly suppresses
  22. L13:1625 weakly suppresses
  23. L9:1368 weakly suppresses
  24. L15:2413 weakly suppresses

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 -1.60 and -1.491 values between -1.49 and -1.382 values between -1.27 and -1.171 values between -1.06 and -0.957 values between -0.95 and -0.847 values between -0.84 and -0.7411 values between -0.74 and -0.6321 values between -0.63 and -0.5227 values between -0.52 and -0.4144 values between -0.41 and -0.30139 values between -0.30 and -0.20461 values between -0.20 and -0.091919 values between -0.09 and 0.02838 values between 0.02 and 0.13225 values between 0.13 and 0.2395 values between 0.23 and 0.3449 values between 0.34 and 0.4520 values between 0.45 and 0.5610 values between 0.56 and 0.678 values between 0.67 and 0.777 values between 0.77 and 0.883 values between 0.88 and 0.993 values between 0.99 and 1.103 values between 1.10 and 1.201 values between 1.20 and 1.312 values between 1.31 and 1.421 values between 1.64 and 1.741 values between 1.74 and 1.851 values between 2.61 and 2.71offnegativepositiveoften

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. const
  2. 我已经
  3. nEnter
  4. 宫
  5. �
  6. изм
  7. 已
  8. 已达
  9. 博物馆
  10. 已经是

Suppresses

  1. 'autres
  2. rade
  3. 更重要的
  4. 焘
  5. cheid
  6. 重要原因
  7. ail
  8. DBNull
  9. 其它
  10. печ