Neuron Explorer

Qwen3-0.6B

'give' or 'gave' before 'rise to'

A neuron in layer 22 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on the word 'give' (or 'gave' or 'gives') when it is followed by 'rise to', as in 'give rise to a disease'. When it was turned up, the model repeated the word 'rise' again and again, which matches what it responds to.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    later, around 1500BC, Austronesian seafarers who entered Near Oceania mixed with diverse groups already there, eventually giving birth to the distinct Lapita people, from which arose Polynesian cultures.⏎

  2. #2 peak

    . A variety of diseases owe their origins to defects in the myosin family of molecular motors: defects in myosins give rise to deafness (VI, VII and others), neurological deficits (myos

  3. #3 peak

    cells of mesenchymal origin. Upon HGF stimulation, MET induces several biological responses that collectively give rise to a program known as invasive growth.⏎Abnormal MET activation in cancer correlates

  4. #4 peak

    BP would integrate these discoveries, he said, and create models to explore how neural circuits are organized, and how they give rise to behaviour and cognition — among the deepest mysteries in neuroscience. Ultimately, said

  5. #5 peak

    ROME X DO? What does Metabolic Syndrome X cause? What are the effects of Metabolic Syndrome X? Over time, normally years, the metabolic imbalances give rise to a greater risk of cardiovascular disease, heart disease, kidney failure, type

  6. #6 peak

    the form of situations and circumstances that may give rise to a seizure.⏎“From here we can then provide advice to patients,

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 need to be able to model a system that is not a single unit. In order to model a system that is not a single unit, I must be able to model something that is a complex and/or a structured and/or a multiple

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

    ave the program a bad reputation that I don't think it really deserved," says Neufeld. "The Vanguard launch came after the success of Sputnik 1 and Sputnik 2,

  2. #2 peak

    commonly used as a model organism<|endoftext|>The Nieuport 80 was a two-seat trainer with one set of controls, developed from the Nieuport 12 and Nieuport 13.⏎The Nieuport 80

  3. #3 peak

    the body uses and stores energy and how the body grows. The thyroid gland makes two forms of thyroid hormone, T4 (thyroxine) and T3 (triiodothyronine).⏎Many babies with congenital hypothyroid

  4. #4 peak

    , a lack of improvement after taking antibiotics for 48 hours⏎Strep throat is caused by infection with a bacterium known as Streptococcus pyogenes, also called group A streptococcus.⏎Streptococ

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. L21:897 suppresses
  2. L19:1335 weakly suppresses
  3. L17:1723 weakly excites
  4. L13:1998 weakly excites
  5. L21:1651 weakly excites
  6. L17:847 weakly suppresses
  7. L20:1040 weakly suppresses
  8. L21:953 weakly excites
  9. L16:1395 weakly suppresses
  10. L16:1892 weakly suppresses
  11. L9:1238 weakly suppresses
  12. L6:2143 weakly excites
  13. L5:1035 weakly excites
  14. L9:298 weakly suppresses
  15. L19:2191 weakly suppresses
  16. L0:634 weakly excites
  17. L19:689 weakly excites
  18. L9:2934 weakly suppresses
  19. L18:1768 weakly suppresses
  20. L9:240 weakly excites
  21. L12:2562 weakly suppresses
  22. L4:2188 weakly suppresses
  23. L9:2611 weakly excites 'is' right after 'there'
  24. L11:150 weakly excites

Feeds later layers

  1. L27:2761 excites
  2. L25:1422 excites
  3. L23:1713 weakly excites
  4. L27:2205 weakly excites
  5. L27:1648 weakly excites
  6. L23:1454 weakly suppresses
  7. L26:1806 weakly excites
  8. L27:2876 weakly excites
  9. L27:285 weakly excites
  10. L27:1708 weakly excites
  11. L23:1850 weakly excites
  12. L23:1168 weakly excites
  13. L25:671 weakly excites
  14. L26:520 weakly suppresses
  15. L25:2310 weakly excites
  16. L25:1543 weakly excites
  17. L24:2415 weakly suppresses
  18. L23:2610 weakly excites
  19. L23:2668 weakly suppresses
  20. L27:1412 weakly excites
  21. L24:415 weakly excites
  22. L23:2195 weakly excites
  23. L25:2366 weakly excites
  24. L26:1101 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 -14.13 and -13.282 values between -13.28 and -12.431 values between -11.58 and -10.741 values between -9.89 and -9.041 values between -9.04 and -8.192 values between -8.19 and -7.355 values between -7.35 and -6.503 values between -6.50 and -5.6510 values between -5.65 and -4.816 values between -4.81 and -3.9615 values between -3.96 and -3.1119 values between -3.11 and -2.2640 values between -2.26 and -1.42196 values between -1.42 and -0.572853 values between -0.57 and 0.28624 values between 0.28 and 1.1362 values between 1.13 and 1.9724 values between 1.97 and 2.8212 values between 2.82 and 3.678 values between 3.67 and 4.525 values between 4.52 and 5.364 values between 5.36 and 6.212 values between 6.21 and 7.063 values between 7.06 and 7.901 values between 7.90 and 8.754 values between 8.75 and 9.601 values between 10.45 and 11.291 values between 12.14 and 12.991 values between 18.07 and 18.921 values between 18.92 and 19.77offnegativepositiveoften

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. rise
  2. 生
  3. 升起
  4. Rise
  5. raise
  6. rise
  7. rising
  8. -rise
  9. emergence
  10. risen

Suppresses

  1. _spaces
  2. .VERTICAL
  3. beeld
  4. 疫情期间
  5. Spaces
  6. 大家都在
  7. 专场
  8. 洮
  9. unate
  10. .NoError