Neuron Explorer

Qwen3-0.6B

number or citation fragment split

A neuron in layer 11 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 spots in citations, page numbers, and view counts where a number or identifier is about to continue, like "U.S. 217" or "Views: 1,130". The steering results do not show a clear theme.

Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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

    Traditional Approaches⏎i. Traditional approach: law of where the injury occurs.⏎ii. Non-intentional Torts⏎1. Alabama Great Southern RR v. Carroll (AL, 1892): P is citizen of AL,

  2. #2 peak

    v. City of Opelika 316 U.S. 584 (1942) and 319 U.S. 103 (1943) raised important issues regarding free exercise of

  3. #3 peak

    . Journal of Marketing, 18(1), 39-53. http://dx.doi.org/10.1080/13527266.2011.62

  4. #4 peak

    :59 PM MDT · Views: 1,130⏎HUMAN EMBRYONIC STEM CELL RE

  5. #5 peak

    igación con células madre embrionarias humanas revela los primeros pasos en el desarrollo humanoContributed by: AdminStaff1 · Views: 1,130⏎Contributed by: AdminStaff1 · April

  6. #6 peak

    Statistics software.⏎The software version addressed is SAS LASR Analytics Server 2.5.⏎SAS In-Memory Statistics Overview<|endoftext|>Thomas Edison Pitch Contest

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 thinking about the old, and I have been thinking about the old, but I have been thinking about the old, and I have been thinking about the old, but I have been thinking about the old, and I have been

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

    set the new global development agenda for the next 15 years. Significant among these are SDG 11 and SDG 16. SDG 11 purposes to “make cities inclusive, safe, resilient and sustainable” and

  2. #2 peak

    30 and No. 3 constructed in 1938 have been abandoned. They were located in Iverson Park. The present system consists of Wells No. 4 through 10, two elevated storage tanks, one ground level

  3. #3 peak

    and the other is big, such as India and Sri Lanka.⏎Figures 2.1 and 2.2 depict two different population pyramids.⏎Figure 2.1 reflects the condition of age-sex pyramid of developing countries like India while

  4. #4 peak

    .⏎Error correcting is defined in 4 level as below.⏎Size of QRcode is defined as version.⏎Version is from 1 to 40.⏎Version 1 is 21*21 matrix. And 4 modules increases whenever

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. L8:350 suppresses blank space before a year
  2. L9:270 suppresses
  3. L9:676 suppresses
  4. L10:404 suppresses
  5. L10:424 suppresses
  6. L6:1244 suppresses
  7. L8:283 suppresses
  8. L9:1958 excites
  9. L8:129 excites
  10. L9:349 suppresses the spot before a year
  11. L8:2006 excites
  12. L8:320 excites
  13. L9:29 excites opening parenthesis before an acronym
  14. L9:2055 suppresses
  15. L10:121 suppresses
  16. L4:1374 suppresses
  17. L6:2785 excites
  18. L8:455 suppresses
  19. L9:831 suppresses
  20. L8:134 excites
  21. L10:1748 excites
  22. L9:401 excites
  23. L10:152 excites
  24. L5:5 suppresses digit after a label like paper or type

Feeds later layers

  1. L13:182 suppresses
  2. L14:670 suppresses
  3. L12:71 suppresses
  4. L14:233 excites
  5. L14:672 suppresses
  6. L13:175 excites
  7. L13:526 suppresses
  8. L15:42 suppresses
  9. L14:53 suppresses
  10. L15:202 suppresses mixed, digits in URLs and comparisons
  11. L16:985 suppresses
  12. L15:1994 weakly suppresses
  13. L12:453 weakly excites
  14. L14:2499 weakly excites
  15. L16:229 weakly suppresses
  16. L14:1696 weakly excites
  17. L20:191 weakly suppresses
  18. L14:453 weakly excites
  19. L14:182 weakly suppresses
  20. L20:252 weakly suppresses
  21. L18:1139 weakly suppresses
  22. L15:141 weakly suppresses
  23. L14:371 weakly excites
  24. L13:739 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.

2 values between -3.79 and -3.651 values between -3.36 and -3.211 values between -2.92 and -2.782 values between -2.78 and -2.632 values between -2.34 and -2.201 values between -2.20 and -2.052 values between -2.05 and -1.902 values between -1.90 and -1.761 values between -1.76 and -1.619 values between -1.61 and -1.473 values between -1.47 and -1.326 values between -1.32 and -1.189 values between -1.18 and -1.038 values between -1.03 and -0.8918 values between -0.89 and -0.7434 values between -0.74 and -0.6065 values between -0.60 and -0.4596 values between -0.45 and -0.31188 values between -0.31 and -0.16795 values between -0.16 and -0.021531 values between -0.02 and 0.13616 values between 0.13 and 0.28305 values between 0.28 and 0.42127 values between 0.42 and 0.5748 values between 0.57 and 0.7117 values between 0.71 and 0.868 values between 0.86 and 1.006 values between 1.00 and 1.151 values between 1.15 and 1.291 values between 1.29 and 1.442 values between 1.44 and 1.581 values between 1.87 and 2.02offnegativepositiveoften

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. rant
  2. din
  3. +s
  4. KN
  5. orama
  6. ED
  7. FN
  8. essian
  9. PD
  10. ronym

Suppresses

  1. cią
  2. 本期
  3. 固然
  4. bedeut
  5. unei
  6. caught
  7. ango
  8. ingleton
  9. mężczy
  10. tę