Neuron Explorer

Qwen3-0.6B

the "10" in DOI prefixes

A neuron in layer 21 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 "0" in the "10." that starts a digital object identifier, the code in academic references like "doi:10.1016/...". It tends to be quiet on other numbers. Turning it up gave unrelated text, so the steering did not reveal more about it.

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

    <|endoftext|>- freely available⏎Sustainability 2013, 5(7), 3006-3008; doi:10.3390/su5073006⏎I

  2. #2 peak

    cript No. AAAJMR-22-82571(R); Published: 29-Nov-2022, DOI:10.35841/ aaajmr-6.11.

  3. #3 peak

    head injury⏎Ann Afr Surg. 2018;15(2):57-61 DOI:http://dx.doi.org/10.4314/aas.v15i2.5⏎©

  4. #4 peak

    calculated that some parts of the moon spend about a quarter of the moon's orbit protected by Earth's magnetosphere (Geophysical Research Letters, DOI: 10.1029/2007GL03050

  5. #5 peak

    ags in the restoration of grassland butterfly communities: a multi-site assessment. Biological Conservation, 155. 50-58. 10.1016/j.biocon.2012.05

  6. #6 peak

    iodiversity and Conservation (2004) 13: 2511. doi:10.1023/B:BIOC.0000048

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: If you are a human, then the number of ways to select a single valid way to choose a set of possible ways to choose a set of possible ways to select a set of possible ways to select a set of possible ways to select a

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

    As were extracted from specimens using TRIzol reagent. Qiagen's miScript System was used to synthesize cDNA and perform qPCR (Figure 1). Endogenous miRNAs are not polyadenylated, therefore during the

  2. #2 peak

    , has been the focus of intense attention in plant science research for its wide-reaching implications to growth and development (Figure 1). At the heart of this clock is a feedback loop of gene expression known as

  3. #3 peak

    PCR product amplification is detected by the level of fluorescence emitted by SYBR Green, a dye which intercalates into double stranded DNA. Specific miRNA prim

  4. #4 peak

    chemical of paraffin-embedded human ovary tumor using 10201-2-AP (ubiquitin antibody) at dilution of 1:50 (under 40x lens).⏎Protein ubiquit

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. L9:1662 excites
  2. L20:96 excites
  3. L3:992 excites
  4. L19:1426 excites
  5. L15:2767 weakly excites
  6. L4:2867 weakly excites
  7. L20:150 weakly excites
  8. L10:378 weakly suppresses
  9. L3:1259 weakly suppresses
  10. L3:1091 weakly excites
  11. L7:399 weakly excites laboratory and research facility terms
  12. L2:94 weakly suppresses
  13. L9:1805 weakly suppresses
  14. L15:1841 weakly excites
  15. L8:2952 weakly excites
  16. L3:37 weakly suppresses
  17. L6:2904 weakly suppresses
  18. L18:2054 weakly suppresses
  19. L5:2834 weakly suppresses
  20. L4:40 weakly suppresses
  21. L8:1834 weakly excites
  22. L0:1263 weakly suppresses
  23. L20:1136 weakly suppresses
  24. L8:2213 weakly excites

Feeds later layers

  1. L22:902 weakly excites
  2. L22:1474 weakly suppresses
  3. L25:671 weakly excites
  4. L23:2061 weakly suppresses
  5. L22:1948 weakly excites
  6. L23:1484 weakly excites
  7. L22:3052 weakly excites
  8. L23:2755 weakly excites
  9. L25:1244 weakly excites
  10. L25:2739 weakly suppresses
  11. L24:1769 weakly suppresses
  12. L24:526 weakly suppresses
  13. L25:2820 weakly suppresses
  14. L22:117 weakly suppresses
  15. L22:2757 weakly excites
  16. L25:2941 weakly excites
  17. L26:2143 weakly excites
  18. L22:2108 weakly excites
  19. L26:966 weakly excites
  20. L22:2140 weakly excites
  21. L24:867 weakly excites
  22. L26:2351 weakly excites
  23. L24:2228 weakly excites
  24. L25:1673 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 -21.14 and -19.931 values between -17.51 and -16.302 values between -13.88 and -12.671 values between -11.46 and -10.251 values between -10.25 and -9.045 values between -7.83 and -6.628 values between -6.62 and -5.415 values between -5.41 and -4.2011 values between -4.20 and -2.9921 values between -2.99 and -1.78172 values between -1.78 and -0.583519 values between -0.58 and 0.64132 values between 0.64 and 1.8518 values between 1.85 and 3.065 values between 3.06 and 4.263 values between 4.26 and 5.472 values between 7.89 and 9.101 values between 26.04 and 27.25offnegativepositiveoften

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. ób
  2. .forName
  3. ода
  4. á
  5. histó
  6. ('${
  7. <![
  8. Про
  9. 作息
  10. vocals

Suppresses

  1. ucks
  2. uzu
  3. asel
  4. Ksi
  5. setDisplay
  6. ardless
  7. thaw
  8. osci
  9. inned
  10. ềm