Neuron Explorer

Qwen3-0.6B

the digit 2 inside citations

A neuron in layer 4 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 reacts to the number 2 when it sits inside a code, reference or scripture number, such as a DOI, a Bible verse or "USB 2.0". It is quiet on the "2" in units like km2. Steering it did not produce a clear theme.

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

    LD, VAT GB238 5548 36<|endoftext|>Medical patients would probably be

  2. #2 peak

    94A21952571. https://dx.doi.org/10.2305/IUCN.UK.2016-2.RLTS.T6794A21

  3. #3 peak

    1 Jewish Holiday Calendar⏎|Biblical Name||Modern Holiday||Hebrew Date||English Date|⏎|Feast of Trumpets (Lev. 23:23–25; Num

  4. #4 peak

    circumstances over which they had no control. A quick check of Jesus’ parables shows the relationships between slaves/servants and wealthy people. (See Matthew 20:1-16, Luke 16:1-13

  5. #5 peak

    .⏎According to the Torah, in Exodus 12:16-20; Leviticus 23:5-8; Numbers 28:16-25

  6. #6 peak

    spend about a quarter of the moon's orbit protected by Earth's magnetosphere (Geophysical Research Letters, DOI: 10.1029/2007GL030507). "

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: no, not a lot. I don't know that the other... maybe not. Maybe not the time for me the space in. There is a lot. So, what? If I can't be, is there a lot of time

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

    |⏎|• Total||1.09 km2 (0.42 sq mi)|⏎|• Density||13,000/km2 (34,000/sq mi)|⏎The Central Business District

  2. #2 peak

    83 sq mi (210 km2) 13%%|⏎|• Density||401/sq mi (155/km2)|⏎|Time zone||UTC−7 (Mountain)|⏎|• Summer

  3. #3 peak

    .6 square miles (250 km2) of it is land and 0.6 square miles (1.6 km2) of it (0.67%) is water.⏎As of the census

  4. #4 peak

    conditions. We compared weight variation and survival between control groups and mixed groups during three years of monitoring. We found both weight loss and high mortality in E. orbicularis of the mixed groups. This study argues for applying a precaution principle and stopping

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. L3:279 suppresses
  2. L3:352 excites
  3. L3:2825 suppresses
  4. L0:418 excites
  5. L1:2536 weakly excites
  6. L1:82 weakly excites
  7. L3:631 weakly suppresses
  8. L0:17 weakly excites
  9. L1:892 weakly excites
  10. L0:949 weakly suppresses
  11. L0:1725 weakly excites
  12. L2:1936 weakly excites
  13. L3:2408 weakly suppresses
  14. L0:2877 weakly suppresses
  15. L3:986 weakly excites
  16. L3:2485 weakly suppresses
  17. L0:21 weakly suppresses last digit of years like 1970
  18. L2:174 weakly excites
  19. L3:861 weakly suppresses
  20. L3:1523 weakly suppresses
  21. L3:343 weakly excites
  22. L3:3031 weakly excites
  23. L2:2366 weakly suppresses
  24. L2:2935 weakly suppresses

Feeds later layers

  1. L17:1412 excites
  2. L16:1220 excites
  3. L17:128 excites
  4. L19:723 excites
  5. L24:1726 excites
  6. L18:122 excites
  7. L19:418 excites
  8. L20:1378 excites
  9. L9:1032 excites
  10. L9:13 excites
  11. L17:840 excites
  12. L19:851 excites comma inside numbers with thousands
  13. L18:2237 excites
  14. L19:258 excites
  15. L10:715 excites
  16. L22:801 excites the digit 1 in numbers
  17. L15:12 excites
  18. L15:1285 excites
  19. L26:1296 excites word-start 'mult' fragment
  20. L12:713 excites
  21. L10:50 excites
  22. L21:1863 excites
  23. L8:20 excites
  24. L17:157 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 -1.11 and -0.961 values between -0.81 and -0.658 values between -0.65 and -0.5038 values between -0.50 and -0.35388 values between -0.35 and -0.201737 values between -0.20 and -0.051490 values between -0.05 and 0.11165 values between 0.11 and 0.2615 values between 0.26 and 0.415 values between 0.41 and 0.564 values between 0.56 and 0.724 values between 0.72 and 0.872 values between 0.87 and 1.026 values between 1.02 and 1.176 values between 1.17 and 1.335 values between 1.33 and 1.485 values between 1.48 and 1.634 values between 1.63 and 1.782 values between 1.78 and 1.946 values between 1.94 and 2.093 values between 2.09 and 2.241 values between 2.24 and 2.392 values between 2.39 and 2.542 values between 2.54 and 2.701 values between 2.70 and 2.853 values between 2.85 and 3.001 values between 3.00 and 3.151 values between 3.46 and 3.611 values between 4.22 and 4.371 values between 4.83 and 4.98offnegativepositiveoften

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. 请点击
  2. ��
  3. peace
  4. authorize
  5. beth
  6. 殓
  7. 瑁
  8. estate
  9. бит
  10. 响

Suppresses

  1. 符号
  2. букв
  3. 形式
  4. ides
  5. fractional
  6. "-"⏎
  7. symbols
  8. earer
  9. 超越
  10. jem