Neuron Explorer

Qwen3-0.6B

'the' after a preposition or verb

A neuron in layer 13 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 word 'the' (or a similar word like 'combined') when it comes right after a preposition or verb, such as 'substitute for the', 'quality of the' or 'accountable for the'. It is waiting for a noun phrase to follow. Turning it up gave loosely connected, wandering text with no clear topic change.

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

    . ©2015 Walmart. All Rights Reserved.⏎Important Walmart Disclaimer: All content provided is for informational purposes only, and is not intended to substitute for the diagnosis, treatment, or advice of a medical professional. Consult your medical professional for

  2. #2 peak

    provides information of a general nature and is designed for educational purposes only. If you have any concerns about your own health or the health of your child, you should always consult with a physician or other health care

  3. #3 peak

    creative fountain design is important. Various types of sprinklers are widely used in waterscapes to generate various water shapes. At present, there are great differences in the quality of domestically produced nozzles. Compared with similar products abroad, the biggest

  4. #4 peak

    2 billion all-time downloads. In the past 12 months, the app was downloaded almost three-quarters of a billion times, which is more than the combined downloads of Facebook, Instagram, and WhatsApp. It is currently in the crossh

  5. #5 peak

    in a locked location separate from the firearm.⏎Child Access Prevention (CAP) laws go one step further by imposing criminal liability to ensure that owners are held accountable for the security of their weapons.⏎The US Congress should pass legislation requiring the safe and secure

  6. #6 peak

    nanobridge brings two worlds together⏎(Nanowerk News) A gadolinium layer of no more than one nanometer in thickness is capable of combining the magnetic world with electronics. In this way, it will be possible to put a

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: I have been writing down the first post: 1.143577550579790543603159033844094571

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

    adult, and is located in the right upper quadrant of the abdominal cavity, just below the diaphragm. It’s functions include:⏎Manufactures half of the body’s cholesterol⏎Production of bile⏎Manufactures other substances including proteins vital to our

  2. #2 peak

    triangle is equal to the sum of the squares of the sides containing the right angle." But Pythagoras had an all-inclusive vision,

  3. #3 peak

    is essential to understand the performance of microorganisms because they play an important role in the day to day lives on earth.⏎Microbiology is the study of the living organisms which are small and cannot be seen with the naked eye. Some of the micro

  4. #4 peak

    a low-lying seamount. Researchers writing in the upcoming Feb. 15 issue of the journal Geochimica et Cosmochimica Acta estimate that the volcanic island will continue to grow, thanks to plate tecton

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. L8:2257 suppresses
  2. L10:669 excites
  3. L10:120 weakly suppresses
  4. L4:2640 weakly suppresses
  5. L12:2287 weakly suppresses
  6. L1:1158 weakly suppresses
  7. L12:2884 weakly suppresses
  8. L11:497 weakly suppresses
  9. L11:2766 weakly suppresses
  10. L12:850 weakly excites
  11. L12:1956 weakly suppresses
  12. L5:1133 weakly excites
  13. L4:654 weakly suppresses
  14. L2:1467 weakly suppresses
  15. L8:2528 weakly excites
  16. L8:1179 weakly suppresses
  17. L5:1882 weakly excites
  18. L10:1290 weakly excites
  19. L11:2465 weakly suppresses
  20. L0:1717 weakly excites
  21. L10:1101 weakly suppresses
  22. L4:390 weakly excites "and" joining abstract nouns
  23. L12:1286 weakly suppresses
  24. L11:2840 weakly excites 'in' after a size measurement

Feeds later layers

  1. L14:727 excites
  2. L17:1959 excites
  3. L18:883 excites
  4. L16:940 excites
  5. L15:2736 excites
  6. L16:1296 excites
  7. L14:2492 excites
  8. L16:601 suppresses
  9. L15:1226 excites
  10. L18:306 excites
  11. L21:1921 excites
  12. L15:1234 excites
  13. L19:566 suppresses
  14. L19:1523 excites
  15. L14:1667 suppresses
  16. L16:307 suppresses
  17. L21:2105 excites
  18. L19:3054 excites
  19. L14:335 weakly suppresses
  20. L15:2496 weakly suppresses
  21. L17:1592 weakly excites
  22. L18:2370 weakly excites
  23. L22:1620 weakly excites "the" after abstract verbs like allows for
  24. L17:430 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.

2 values between -1.81 and -1.652 values between -1.65 and -1.493 values between -1.49 and -1.332 values between -1.33 and -1.175 values between -1.17 and -1.0111 values between -1.01 and -0.8513 values between -0.85 and -0.6932 values between -0.69 and -0.5362 values between -0.53 and -0.37164 values between -0.37 and -0.21651 values between -0.21 and -0.052018 values between -0.05 and 0.11579 values between 0.11 and 0.27225 values between 0.27 and 0.4357 values between 0.43 and 0.5918 values between 0.59 and 0.7513 values between 0.75 and 0.9112 values between 0.91 and 1.073 values between 1.07 and 1.235 values between 1.23 and 1.396 values between 1.39 and 1.551 values between 1.55 and 1.711 values between 1.71 and 1.873 values between 1.87 and 2.032 values between 2.03 and 2.193 values between 2.19 and 2.352 values between 2.35 and 2.512 values between 2.51 and 2.672 values between 2.83 and 2.991 values between 2.99 and 3.151 values between 3.15 and 3.311 values between 3.31 and 3.472 values between 3.47 and 3.631 values between 3.63 and 3.791 values between 3.95 and 4.111 values between 4.11 and 4.271 values between 4.43 and 4.59offnegativepositiveoften

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. של
  2. >&
  3. >>&
  4. management
  5. ind
  6. uzu
  7. ��
  8. hei
  9. еств
  10. activities

Suppresses

  1. ː
  2. intree
  3. caster
  4. ||⏎
  5. tố
  6. 做到
  7. 做的
  8. 听得
  9. 作了
  10. 说得