Neuron Explorer

Qwen3-0.6B

"the" after time or place prepositions

A neuron in layer 20 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" when it follows a preposition such as "at", "from", or "over" in phrases like "at the same time", "from the perspective of" or "over the past decades". Its activations are weak compared with the other neurons here. Turning it down made the model output "the the the" repeatedly, so the outputs do not show the trigger clearly.

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

    by 50% before 2030.⏎This commitment isn’t just good for the planet—it also comes with considerable financial benefits. Numerous studies over the past three decades have proven the value of green building certifications, with a rental premium

  2. #2 peak

    future that is not necessarily entangled with typical database concerns. There are numerous graph technologies that each have their respective benefits and drawbacks. Leveraging the right technology at the right time is required for effective graph computing.⏎Structure: Modeling Real-World Sc

  3. #3 peak

    up for outputs of up to 75 watts, but by swapping out parts for ones with higher power ratings you can power larger appliances or multiple smaller appliances at the same time. This lets you build a system that fits your energy needs and your

  4. #4 peak

    future directions for the field. Suitable for introductory courses and ideal for students new to the subject, this concise and lively study reconsiders the history of science from the perspective of world and comparative history.<|endoftext|>A marker specifies a buffer and a position

  5. #5 peak

    to eat more fruits and vegetables and less processed foods combined with Americans’ favorite past-time of adopting restrictive, hard-to-maintain, weight-loss inducing diets at the turn of the New Year, creates a perfect storm for a surge in interest and

  6. #6 peak

    2015, 54 of Atlanta Public Schools’ 81 schools implemented CEP<|endoftext|>I’ve been a longtime fan of Understanding Science from the University of California Museum of Paleontology at Berkeley. It’s a comprehensive resource for

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 the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the the

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

    in the recent past; we have seen several instances of Governors holding Bills passed by the State Legislatures for inordinate amount of time, without deciding one way or the other. In some cases, even when ex-facie the Bill has to be

  2. #2 peak

    mind learns and how this information can be applied both in academics and in life.⏎Bain, Ken. What the Best College Students Do. Cambridge, MA: The Belnap Press, 2012. Print.⏎Using both quantitative research

  3. #3 peak

    Solving⏎Tier 2—Assessment and Problem Solving⏎We can no longer implement instruction we believe to be great and cross our fingers and hope for the best. We must identify appropriate strategies the way a physician determines an appropriate prescription.

  4. #4 peak

    is then transferred to the trip coil which operates to trip the circuit breaker.⏎Email Based, Online Homework Assignment Help in Classification Of Relays⏎Transtutors is the best place to get answers to all your doubts regarding classification of relays with examples

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. L19:2283 strongly suppresses
  2. L18:861 strongly suppresses
  3. L19:1023 strongly excites
  4. L19:644 strongly excites
  5. L19:1180 strongly suppresses
  6. L18:1359 excites
  7. L11:223 excites
  8. L16:1320 suppresses
  9. L11:1168 excites
  10. L13:44 suppresses inside web addresses and citations
  11. L10:155 excites
  12. L17:30 suppresses
  13. L15:184 excites
  14. L18:1041 suppresses
  15. L18:178 excites
  16. L18:1909 suppresses
  17. L7:92 excites
  18. L18:1513 suppresses
  19. L17:372 suppresses
  20. L3:344 excites
  21. L10:1644 excites
  22. L17:1056 excites
  23. L12:158 suppresses
  24. L18:635 excites

Feeds later layers

  1. L24:153 strongly suppresses
  2. L21:614 strongly excites
  3. L21:592 excites
  4. L25:390 excites
  5. L21:675 suppresses
  6. L25:544 suppresses
  7. L24:327 excites
  8. L25:284 suppresses
  9. L23:79 suppresses
  10. L21:310 suppresses
  11. L27:193 suppresses
  12. L21:547 excites
  13. L24:261 suppresses
  14. L26:1941 excites
  15. L24:918 suppresses
  16. L21:2342 suppresses
  17. L27:173 excites
  18. L27:421 excites
  19. L22:171 suppresses
  20. L27:797 suppresses
  21. L23:392 suppresses
  22. L27:663 suppresses
  23. L22:2302 suppresses
  24. L26:213 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 -15.98 and -15.331 values between -6.85 and -6.201 values between -6.20 and -5.554 values between -4.25 and -3.593 values between -3.59 and -2.943 values between -2.94 and -2.2923 values between -2.29 and -1.64718 values between -1.64 and -0.991514 values between -0.99 and -0.33952 values between -0.33 and 0.32233 values between 0.32 and 0.97106 values between 0.97 and 1.6280 values between 1.62 and 2.2758 values between 2.27 and 2.9241 values between 2.92 and 3.5844 values between 3.58 and 4.2338 values between 4.23 and 4.8819 values between 4.88 and 5.5323 values between 5.53 and 6.1818 values between 6.18 and 6.8414 values between 6.84 and 7.491 values between 7.49 and 8.147 values between 8.14 and 8.793 values between 8.79 and 9.443 values between 9.44 and 10.09offnegativepositiveoften

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. 好的
  2. own
  3. same
  4. 该
  5. 該
  6. 自己
  7. same
  8. oretical
  9. Crow
  10. 平

Suppresses

  1. кра
  2. EndInit
  3. очно
  4. ıy
  5. téc
  6. lijk
  7. (errno
  8. ело
  9. سوء
  10. ела