Neuron Explorer

Qwen3-0.6B

space token right after a period

A neuron in layer 0 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 blank space that appears right after a full stop, mostly in citations, reference lists and cut-up text. It is a weak and fairly unclear signal. When it was turned up, the replies did not show a clear shared 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

    found to be “obstructive.”<|endoftext|>Rural earthen roads impact assessment in Phewa watershed, Western region, Nepal⏎© The Author(s). 2016⏎Received: 23 November 2015

  2. #2 peak

    . 4- The subject becomes object and use (by) before it.⏎SubjectObject

  3. #3 peak

    . 2005. Edge Effects in Line Intersect Sampling With. Journal Agricultural

  4. #4 peak

    . 29:1–6)||Rosh Hashanah (head<|endoftext|>T

  5. #5 peak

    . 25:24–25.)⏎The law leads us back home;

  6. #6 peak

    . ¼ mile away). Touch for a list and map of all markers in Georgetown.⏎

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: to get the to get the to get the to get the to get the to get the to get the to get the to get the I have been thinking about lately: to get the to

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

    .…It's less powerful than even a small…watch battery.…For processing and mixing

  2. #2 peak

    and his followers, social capital is a key component to building and maintaining democracy.⏎Communitarians seek to bolster social capital and the institutions of civil society. . . .⏎Central to many communitarians' philosophy is the concept of positive rights;

  3. #3 peak

    . Armed with the knowledge of where<|endoftext|>Edited by William Byron Forbush This is a book that will never die — one of the great English classics. . .

  4. #4 peak

    next-generation system was twice as much as it originally calculated, so it was forced to shut down the Arizona pilot and lay off nearly half of its staff.⏎. . .⏎So far so good. Those who think the algae are done are in for

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

None: this is the first layer.

Feeds later layers

  1. L1:130 excites
  2. L1:441 weakly excites
  3. L4:295 weakly excites
  4. L16:432 weakly excites period after month or initial
  5. L13:148 weakly excites
  6. L1:1071 weakly excites period inside abbreviations like i.e. and e.g
  7. L1:97 weakly excites full stop ending a heading or label
  8. L2:156 weakly excites
  9. L2:1243 weakly suppresses
  10. L1:1054 weakly excites "especially" after a comma
  11. L18:713 weakly excites
  12. L13:147 weakly excites
  13. L16:1802 weakly suppresses
  14. L15:313 weakly excites decimal point inside numbers
  15. L7:1760 weakly suppresses
  16. L1:448 weakly suppresses
  17. L3:595 weakly excites
  18. L1:1211 weakly suppresses
  19. L7:1219 weakly suppresses
  20. L3:754 weakly suppresses
  21. L17:2547 weakly suppresses
  22. L1:443 weakly suppresses
  23. L5:1329 weakly excites
  24. L2:1205 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.

5 values between -0.33 and -0.3013 values between -0.30 and -0.2819 values between -0.28 and -0.2518 values between -0.25 and -0.2313 values between -0.23 and -0.2011 values between -0.20 and -0.1814 values between -0.18 and -0.1547 values between -0.15 and -0.13111 values between -0.13 and -0.10110 values between -0.10 and -0.08168 values between -0.08 and -0.05349 values between -0.05 and -0.03644 values between -0.03 and -0.01824 values between -0.01 and 0.02646 values between 0.02 and 0.05454 values between 0.05 and 0.07279 values between 0.07 and 0.10115 values between 0.10 and 0.1236 values between 0.12 and 0.1414 values between 0.14 and 0.175 values between 0.17 and 0.195 values between 0.19 and 0.221 values between 0.22 and 0.241 values between 0.24 and 0.271 values between 0.32 and 0.341 values between 0.39 and 0.421 values between 0.44 and 0.471 values between 0.47 and 0.491 values between 0.49 and 0.521 values between 0.64 and 0.67offnegativepositiveoften

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. œur
  2. ctest
  3. ives
  4. пуска
  5. come
  6. levant
  7. 不同
  8. ugins
  9. ʎ
  10. 汐

Suppresses

  1. sov
  2. Kont
  3. ItemType
  4. 穑
  5. enschaft
  6. .Annotations
  7. ronics
  8. urg
  9. `%
  10. `_