Neuron Explorer

Qwen3-0.6B

digits inside citation numbers

A neuron in layer 21 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on single digits inside reference details like volume numbers, page ranges, dates and DOI codes, such as the 4 in 2010/04/19 or the 5 in 524. Steering was mixed, with number sequences showing up when it was turned up.

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

    + 1 = 2. We then get the answer 524 Can we simplify? Yes, that's possible. 524 = 512.⏎The answer is 214 + 314 =

  2. #2 peak

    foods, so follow these tips to shake your salt habit:⏎*http://www.washingtonpost.com/wp-dyn/content/article/2010/04/19/AR

  3. #3 peak

    and articles⏎James Behuniak Jr (2011). Naturalizing Mencius. Philosophy East and West 61 (3):492-515.⏎David E. Soles (1999).

  4. #4 peak

    , CT scan, rule, minor head injury⏎Ann Afr Surg. 2018;15(2):57-61 DOI:http://dx.doi.org/10.43

  5. #5 peak

    , and it can exceed MRI when assessing for partial tendon tears because of its ability to perform dynamic testing. (Eur Radiol 2012;22:1140.) Ultrasonography also has been shown to be

  6. #6 peak

    native and non‐native plants to disturbance and herbivory in mountain environments. Journal of Biogeography, 48(7), 1594-1605 https://doi.org/10.111

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 to do some changes to the way the database is structured to make it easier for the users to use. To do so, I will make some changes to the database structure, and I will make some changes to the database to be

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

    with a while loop:⏎given_arr = [1, 2, 3, 4, 5] i = len(given_arr) - 1 while(i >= 0): print(given_arr[i]) i = i -

  2. #2 peak

    that have a high pH value (10 to 14), which can be quite caustic. The use of acidic products

  3. #3 peak

    it that way. That just means that I have to make sure that I use gravity as a negative number, since gravity always acts down.⏎vf = vi + at⏎= 56.3m/s + (-9.81m

  4. #4 peak

    perpendicular.⏎Let's see how to identify parallel and perpendicular lines.⏎Let's take some examples to understand this.⏎Considering two equations:⏎y = 2x +3 and y = 2x+5⏎On comparing y = 2x

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. L19:232 suppresses
  2. L19:1549 excites
  3. L18:100 weakly suppresses
  4. L18:2078 weakly excites
  5. L20:737 weakly suppresses
  6. L16:743 weakly excites
  7. L19:2083 weakly excites
  8. L20:2813 weakly excites
  9. L8:525 weakly excites
  10. L11:249 weakly excites
  11. L12:96 weakly suppresses
  12. L17:1152 weakly excites
  13. L16:325 weakly suppresses
  14. L18:1132 weakly suppresses
  15. L9:31 weakly excites
  16. L16:2660 weakly excites
  17. L12:2336 weakly suppresses
  18. L20:80 weakly excites
  19. L17:2110 weakly suppresses
  20. L18:2056 weakly excites
  21. L0:1110 weakly suppresses
  22. L3:1550 weakly suppresses
  23. L0:1980 weakly excites
  24. L20:156 weakly suppresses

Feeds later layers

  1. L22:50 weakly excites
  2. L23:559 weakly excites
  3. L26:403 weakly excites word starts like cr, fr, kr
  4. L27:744 weakly excites
  5. L26:235 weakly excites
  6. L23:1358 weakly excites
  7. L25:1030 weakly excites
  8. L23:1237 weakly excites
  9. L24:1695 weakly suppresses
  10. L24:1542 weakly excites
  11. L23:507 weakly excites
  12. L24:173 weakly excites
  13. L27:2198 weakly excites
  14. L24:673 weakly excites
  15. L26:32 weakly suppresses
  16. L25:1577 weakly excites
  17. L27:789 weakly excites
  18. L24:1394 weakly excites
  19. L22:2439 weakly excites
  20. L24:1878 weakly excites
  21. L26:316 weakly excites
  22. L23:1232 weakly suppresses
  23. L24:539 weakly excites
  24. L25:2689 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 -7.40 and -6.951 values between -6.95 and -6.511 values between -6.51 and -6.062 values between -6.06 and -5.611 values between -5.61 and -5.161 values between -5.16 and -4.711 values between -4.71 and -4.271 values between -4.27 and -3.823 values between -3.82 and -3.371 values between -3.37 and -2.923 values between -2.92 and -2.4711 values between -2.47 and -2.0319 values between -2.03 and -1.5825 values between -1.58 and -1.1361 values between -1.13 and -0.68686 values between -0.68 and -0.232410 values between -0.23 and 0.22423 values between 0.22 and 0.66121 values between 0.66 and 1.1155 values between 1.11 and 1.5621 values between 1.56 and 2.0118 values between 2.01 and 2.4615 values between 2.46 and 2.915 values between 2.91 and 3.3510 values between 3.35 and 3.802 values between 3.80 and 4.251 values between 4.25 and 4.701 values between 4.70 and 5.152 values between 5.15 and 5.593 values between 5.59 and 6.042 values between 9.18 and 9.631 values between 10.08 and 10.52offnegativepositiveoften

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. (
  2. send
  3. .Http
  4. (Http
  5. osen
  6. 'https
  7. "https
  8. [[
  9. Http
  10. ió

Suppresses

  1. ניים
  2. тра
  3. emode
  4. rians
  5. ration
  6. ăn
  7. Directive
  8. oji
  9. 简化
  10. ☟