Neuron Explorer

Qwen3-0.6B

double bar table cell separators

A neuron in layer 6 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 responds to the doubled vertical bar "||" that separates cells in flattened web tables, such as a label followed by its value. Turning it up produced unrelated rambling text, so the steering does not confirm the pattern.

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

    ,4) and (2,3). Then the table would look like this:⏎1 2 3 4⏎1 T T F T⏎2 T T T F⏎3 F T T F⏎4 T F F T<|endoftext|>

  2. #2 peak

    100%|⏎|20 years||100-170 beats per minute||200 beats per minute|⏎|30 years||95-162 beats per minute||190 beats per minute

  3. #3 peak

    60 direct from Amazon (22% off) Amazon page|⏎|Call number||QC7.H62 2005|⏎|External links||This entry has no external links. Add one.|⏎|Through your library||

  4. #4 peak

    |⏎|SPRING 2016||ARAB 200 AND ARAB 300|⏎|FALL 2016||ARAB 100 AND ARAB 210⏎This projection

  5. #5 peak

    11)|| 0.805⏎very high · 43rd⏎|Currency||Euro (€) (EUR)|⏎|Time zone||EET (UTC+2)|⏎|-||Summer (DST)||E

  6. #6 peak

    is a full breakdown.⏎15 Hours of Mathematics Instruction⏎|Day 1 – Algebra 1||3 hours||–|⏎|Day 2 – Geometry||3 hours||Algebra 1 Quiz|⏎|Day 3 – Arithmetic||

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: The math in the curriculum. The set S = {0, 1, a, b} The set S = {0, 1, a, b} So, the set S is: S = {0, 1

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

    VE Articles!⏎Assessment of Social Cognition in Non-human Primates Using a Network of Computerized Automated Learning Device (ALDM) Test Systems⏎Institutions: Aix-Marseille University.⏎Fag

  2. #2 peak

    amplification and RNA-Seq.⏎19 Related JoVE Articles!⏎Detection of Rare Genomic Variants from Pooled Sequencing Using SPLINTER⏎Institutions: Washington University School of Medicine, Washington University School of Medicine, Washington University School of

  3. #3 peak

    Radio-telemetric System to Monitor Cardiovascular Function in Rats with Spinal Cord Transection and Embryonic Neural Stem Cell Grafts⏎Institutions: Drexel University College of Medicine, Heidelberg University Hospital, Veterans Administration Medical

  4. #4 peak

    and high-dose irradiation of a precisely defined target volume (tumor). It is usually performed in a session lasting about 30 minutes. Radiosurgery involves treating tumors in the head or body. The tumor cells

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. L4:1647 weakly suppresses
  2. L0:2932 weakly excites
  3. L5:3023 weakly excites
  4. L0:2576 weakly excites
  5. L5:2603 weakly suppresses
  6. L5:392 weakly suppresses
  7. L1:2329 weakly excites
  8. L2:977 weakly excites
  9. L2:1657 weakly excites
  10. L0:1347 weakly excites
  11. L2:237 weakly excites
  12. L5:2154 weakly excites
  13. L2:2635 weakly excites
  14. L1:2312 weakly excites
  15. L1:2683 weakly excites
  16. L1:1831 weakly suppresses
  17. L4:14 weakly excites
  18. L2:3003 weakly suppresses
  19. L4:637 weakly excites
  20. L0:1286 weakly excites
  21. L2:908 weakly suppresses
  22. L5:2368 weakly suppresses smell and taste words in health text
  23. L2:484 weakly excites
  24. L5:533 weakly excites

Feeds later layers

  1. L20:1521 weakly suppresses
  2. L14:97 weakly excites
  3. L17:541 weakly suppresses
  4. L15:959 weakly suppresses unclear: sentence ends in how-to text
  5. L18:2607 weakly suppresses
  6. L15:2208 weakly suppresses
  7. L13:812 weakly suppresses
  8. L8:2303 weakly suppresses
  9. L22:794 weakly suppresses
  10. L23:2523 weakly suppresses
  11. L16:1794 weakly excites
  12. L7:1717 weakly suppresses
  13. L17:1801 weakly suppresses
  14. L11:37 weakly suppresses
  15. L7:2702 weakly excites
  16. L12:338 weakly excites
  17. L21:2703 weakly excites
  18. L15:2734 weakly excites
  19. L21:1458 weakly suppresses
  20. L22:2357 weakly excites
  21. L9:1238 weakly excites
  22. L8:1187 weakly excites
  23. L7:1220 weakly excites
  24. L11:1473 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 -0.57 and -0.542 values between -0.54 and -0.516 values between -0.51 and -0.485 values between -0.48 and -0.459 values between -0.45 and -0.424 values between -0.42 and -0.3913 values between -0.39 and -0.3619 values between -0.36 and -0.3337 values between -0.33 and -0.3040 values between -0.30 and -0.2754 values between -0.27 and -0.2496 values between -0.24 and -0.21115 values between -0.21 and -0.18162 values between -0.18 and -0.15190 values between -0.15 and -0.12246 values between -0.12 and -0.09311 values between -0.09 and -0.06344 values between -0.06 and -0.03407 values between -0.03 and -0.01404 values between -0.01 and 0.03309 values between 0.03 and 0.06268 values between 0.06 and 0.09251 values between 0.09 and 0.11155 values between 0.11 and 0.14122 values between 0.14 and 0.17108 values between 0.17 and 0.2070 values between 0.20 and 0.2344 values between 0.23 and 0.2635 values between 0.26 and 0.2927 values between 0.29 and 0.3216 values between 0.32 and 0.359 values between 0.35 and 0.389 values between 0.38 and 0.414 values between 0.41 and 0.4410 values between 0.44 and 0.472 values between 0.47 and 0.502 values between 0.50 and 0.531 values between 0.53 and 0.561 values between 0.59 and 0.62offnegativepositiveoften

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. 書き
  2. ophys
  3. aciones
  4. ény
  5. ujet
  6. templ
  7. apphire
  8. álido
  9. 惚
  10. ẫu

Suppresses

  1. ока
  2. wick
  3. ()</
  4. role
  5. ch
  6. ki
  7. 肩
  8. ::
  9. ordinates
  10. ://%