Neuron Explorer

Qwen3-0.6B

'poly' word-start fragment

A neuron in layer 24 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

Fires on the start of chemistry and materials words that begin with 'poly', like polyethylene, polymer and polyol. Turning it up made the model produce strings of digits and odd repeats, so the steering result does not clearly reflect this 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

    ubiquitylation system.⏎Ubiquitin can be ubiquitinated on seven lysine (Lys) residues or on the N-terminus, leading to polyubiquitin chains that can encompass complex topologies. Alternatively, or in addition

  2. #2 peak

    This is a question that many consumers have when it comes to choosing plastic bottles for their food and beverage products.⏎PET, or polyethylene terephthalate, is a type of plastic that is commonly used

  3. #3 peak

    . The resulting sensor array then was integrated into a microfluidic chip made of poly(dimethylsiloxane), a standard material used to make lab-on-a-ch

  4. #4 peak

    qPCR (Figure 1). Endogenous miRNAs are not polyadenylated, therefore during the reverse transcription process, a poly(A) polymerase polyadenylates the miRNA. The miRNA is used as a template to

  5. #5 peak

    RNA constitutes approximately 90%. It is challenging to enrich messenger RNA from a metagenomic microbial RNA sample because most prokaryotic mRNA species lack stable poly

  6. #6 peak

    carbyne (about 0.71 nm). Our approach of HV high temperature annealing has a great advantage over filling SWCNT/DWCNT by poly

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: once in a while, we get a moment to once again get at once again once again once again at once again once again once again once again once again at once again once again at once again once again once again once again once again once

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

    use, and are used raw. The flowers contain a volatile oil made of a variety of amino acids and other substances, including borneol, camphor, adenine, and small amounts of vitamin B1.⏎In traditional Chinese medicine, ch

  2. #2 peak

    by using chemicals to kill germs on surfaces or objects. According to the CDC, several types of disinfectants are effective against the flu including chlorine bleach, hydrogen peroxide, and quaternary ammonium-based cleaners. No matter which product

  3. #3 peak

    and perirectum also are affected. Signs of infection, including warmth and swelling, may be absent.⏎Neutropenia also suggests the following possibilities:⏎Aplastic anemia reduces the bone marrow's ability to produce white blood cells.⏎Auto

  4. #4 peak

    :⏎|Vitamin B12 Need||94%||Confirm|⏎|Lupus (SLE)||2%||Ruled out|⏎|Aplastic Anemia||1%||Ruled out|⏎|Fanconi Anemia

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. L22:819 suppresses
  2. L23:389 suppresses
  3. L20:2701 excites
  4. L1:180 excites
  5. L22:2552 weakly suppresses
  6. L23:2273 weakly suppresses
  7. L16:2589 weakly excites
  8. L0:356 weakly excites
  9. L21:1590 weakly excites
  10. L7:754 weakly excites
  11. L20:3034 weakly suppresses
  12. L1:363 weakly excites
  13. L2:1207 weakly excites
  14. L10:2671 weakly suppresses
  15. L5:942 weakly excites
  16. L11:2432 weakly excites
  17. L22:2483 weakly excites
  18. L7:287 weakly suppresses
  19. L20:2513 weakly excites
  20. L13:937 weakly suppresses
  21. L5:2695 weakly excites
  22. L8:1367 weakly excites
  23. L10:2954 weakly excites
  24. L5:135 weakly suppresses

Feeds later layers

  1. L25:1749 weakly excites
  2. L25:1700 weakly excites
  3. L25:759 weakly excites
  4. L25:1524 weakly excites
  5. L27:1247 weakly excites
  6. L27:991 weakly suppresses
  7. L27:2747 weakly suppresses
  8. L26:2062 weakly suppresses
  9. L25:605 weakly suppresses
  10. L25:1500 weakly suppresses
  11. L27:1252 weakly suppresses
  12. L25:1058 weakly excites
  13. L27:2846 weakly excites
  14. L26:2836 weakly suppresses
  15. L26:644 weakly suppresses
  16. L27:1571 weakly suppresses
  17. L26:2075 weakly suppresses
  18. L25:228 weakly suppresses
  19. L25:1354 weakly excites
  20. L25:2520 weakly suppresses
  21. L25:2440 weakly excites
  22. L27:887 weakly suppresses
  23. L27:485 weakly excites
  24. L27:519 weakly 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 -33.47 and -32.021 values between -21.90 and -20.461 values between -19.01 and -17.571 values between -16.12 and -14.682 values between -13.23 and -11.782 values between -11.78 and -10.341 values between -10.34 and -8.892 values between -8.89 and -7.451 values between -7.45 and -6.003 values between -6.00 and -4.564 values between -4.56 and -3.118 values between -3.11 and -1.66389 values between -1.66 and -0.223465 values between -0.22 and 1.2317 values between 1.23 and 2.672 values between 2.67 and 4.122 values between 4.12 and 5.571 values between 5.57 and 7.011 values between 7.01 and 8.462 values between 8.46 and 9.901 values between 18.58 and 20.021 values between 22.91 and 24.36offnegativepositiveoften

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. 解放
  2. ac
  3. utt
  4. ad
  5. AD
  6. aint
  7. und
  8. adol
  9. Submission
  10. .toHexString

Suppresses

  1. reap
  2. _SI
  3. iability
  4. LAG
  5. misc
  6. już
  7. Mis
  8. vero
  9. .mi
  10. lã