Neuron Explorer

Qwen3-0.6B

"especially" after a comma

A neuron in layer 1 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 word "especially" when it comes after a comma and adds a special case to a sentence, like "useful, especially to ward off". It is pushed strongly negative on certain punctuation, such as a comma or dash after a name or quote. Turning it up or down did not give a clear shift in the writing.

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

    in some cases, glass bottles offer the better choice, especially for certain types of beverages and storage circumstances. The difference lies with the permeability

  2. #2 peak

    on purees and become picky eaters, especially if they get used to extremely smooth purees. However, you can avoid this

  3. #3 peak

    , UGM, Saturday (31/10).⏎According to Hamidah, eating natural fruits is very useful, especially to ward off various diseases that come with age. "If we go back to

  4. #4 peak

    still appropriate that certain times be set aside for a particular focus on repentance.⏎In much the same way, we celebrate the resurrection of Christ each Sunday, but especially at Easter; and while we should always thank God for the Incarnation, we

  5. #5 peak

    , this can make telling their true age difficult, especially when dealing with strays that somehow warmed their way into our hearts and home.

  6. #6 peak

    subject.⏎• Keep your tiles clean. A clean tile won't have a film of accumulated grinding dirt on the surface.⏎• Avoid the use of unnecessary chemicals, especially acid based ones and the use of hard cleaning implements such as steel wool. These

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 of the subject of the person who is the person who is the person. It is the person. It is the person. It is the person. It is the person. It is the person. It is the person

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

    BALANCAN, MEXICO (EFE).- The finding of 23 archaeological pieces from about 600 BC has

  2. #2 peak

    'S', then click at the end of the word and type<|endoftext|>Mallorca’s popular tourist railway, the ‘Orange Express,’ has entered its 90th year of electrification. The Palma –

  3. #3 peak

    achay-Cherkessia, Russia, near the border of Georgia.⏎Elbrus’ earliest name was Strobilus, Latin for ‘pine cone,’ from the ancient Greek strobilos, meaning ‘a twisted object’, a botanical

  4. #4 peak

    of the difference between 'wearing' and 'scratching.' Wearing is where the glazed surface is thinning and eventually actually work through.

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. L0:178 suppresses
  2. L0:1224 suppresses
  3. L0:215 weakly suppresses
  4. L0:348 weakly excites space token right after a period
  5. L0:1301 weakly suppresses
  6. L0:456 weakly excites
  7. L0:249 weakly excites ordinal suffix "nd" after numbers
  8. L0:319 weakly excites blank space after a label colon
  9. L0:1518 weakly excites
  10. L0:1794 weakly suppresses
  11. L0:2838 weakly suppresses
  12. L0:276 weakly suppresses
  13. L0:899 weakly suppresses
  14. L0:847 weakly excites
  15. L0:2659 weakly excites
  16. L0:622 weakly excites
  17. L0:1276 weakly suppresses
  18. L0:36 weakly excites
  19. L0:2692 weakly suppresses
  20. L0:887 weakly suppresses
  21. L0:1483 weakly excites
  22. L0:414 weakly suppresses blank after a dash in number ranges
  23. L0:1406 weakly excites
  24. L0:585 weakly excites

Feeds later layers

  1. L2:1142 weakly suppresses
  2. L26:2143 weakly suppresses
  3. L25:342 weakly suppresses
  4. L2:986 weakly suppresses
  5. L2:1837 weakly suppresses
  6. L10:2191 weakly excites
  7. L4:2018 weakly excites
  8. L5:1299 weakly excites
  9. L20:603 weakly suppresses
  10. L4:2260 weakly suppresses
  11. L26:1443 weakly suppresses
  12. L27:492 weakly suppresses
  13. L17:277 weakly excites
  14. L2:1974 weakly excites
  15. L20:1422 weakly suppresses
  16. L17:2979 weakly suppresses
  17. L9:334 weakly suppresses
  18. L13:1796 weakly excites
  19. L17:1151 weakly excites
  20. L18:655 weakly suppresses
  21. L26:2115 weakly suppresses
  22. L21:2280 weakly excites
  23. L16:2633 weakly suppresses
  24. L11:2156 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.

1 values between -1.87 and -1.771 values between -1.77 and -1.672 values between -1.07 and -0.971 values between -0.57 and -0.473 values between -0.47 and -0.3713 values between -0.37 and -0.26118 values between -0.26 and -0.16625 values between -0.16 and -0.061914 values between -0.06 and 0.041029 values between 0.04 and 0.14149 values between 0.14 and 0.2420 values between 0.24 and 0.3414 values between 0.34 and 0.4410 values between 0.44 and 0.545 values between 0.54 and 0.641 values between 0.94 and 1.041 values between 1.75 and 1.851 values between 2.05 and 2.15offnegativepositiveoften

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. yen
  2. self
  3. yre
  4. eyond
  5. glomer
  6. crease
  7. ramid
  8. yü
  9. iere
  10. uridad

Suppresses

  1. 康复
  2. Чт
  3. 万美元
  4. устройств
  5. 捐献
  6. 选秀
  7. (es
  8. одерж
  9. 습
  10. ám