Neuron Explorer

Qwen3-0.6B

'not' or 'only' setting up a contrast

A neuron in layer 7 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 lights up on words like "not" or "only" that come right after a verb such as "is" or "are" and set up a contrast, as in "they are not effective" or "though not exclusively". Its negative side fires on words like "typically" and "also" in similar spots. Turning it up made the model write rambling, repetitive text that did not clearly relate to the topic.

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

    source to generate energy. This is known as the Warburg Effect.⏎While metabolic changes are an important feature in the transformation of normal cells into cancer cells they are not now thought to be cancer’s primary cause. Despite this, metabolic changes remain an

  2. #2 peak

    899 and has been used with success in chemical, ceramic, carbon allotropy, steel/alloy, composite materials and plastic industries for decades, it was only in late 1980′s that its commercial benefits became available to

  3. #3 peak

    the novel which is also an alternative history–this one, The Grasshopper Lies Heavy, portraying a world in which the Allies won World War II, though not exactly along the lines of history as we know it. Despite the book being banned

  4. #4 peak

    authorities were suspicious, and added up the total of his deposits. They found that he had deposited $162,<|endoftext|>An intelligent and impassioned, albeit not

  5. #5 peak

    host of health benefits, including weight loss, improved lung function, lower risk of stroke, cancer and heart disease.⏎But just because the humble apple is ubiquitous doesn't mean it's without some mystery as well. Read on to learn more about the

  6. #6 peak

    Until now, semi-conductors have been the main tool for waste heat recovery, but they are not effective when temperatures drop below 100 degrees Celsius. Semi-conductors

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 am not a fan of the words. I am not a fan of the words. So, the words are: **The words are:** **"I am not a fan of the words"** I am not a fan of

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

    . The Stabilizer level of swimming pool water must be maintained at 30-50 ppm. The Stabilizer level is important, but is typically only tested at the beginning of the swimming season when a pool is opened. The

  2. #2 peak

    -related hormones, but research shows it also improves stamina and relieves fatigue. Eleuthero, often called Siberian g

  3. #3 peak

    it's driven by "thermal and geophysical disturbances in the inner Earth", or links between tectonic plates and the climate.⏎But another is that the Earth passes through the plane of the Milky Way galaxy every 3

  4. #4 peak

    to 30 grams of protein at each meal to optimize muscle protein synthesis. Though most people eat a combination of protein-containing foods at each meal, this infographic shows what 25 grams of protein from different sources looks like.⏎Researchers in Ireland

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. L0:914 weakly suppresses
  2. L5:1434 weakly suppresses
  3. L1:1018 weakly suppresses
  4. L5:2324 weakly excites
  5. L3:2698 weakly suppresses
  6. L1:1806 weakly excites
  7. L6:824 weakly excites
  8. L3:1410 weakly excites
  9. L6:1917 weakly excites
  10. L4:2090 weakly excites
  11. L2:1054 weakly excites
  12. L6:1054 weakly excites
  13. L2:1331 weakly excites
  14. L1:2294 weakly excites
  15. L6:2425 weakly suppresses
  16. L1:731 weakly suppresses
  17. L1:1947 weakly excites
  18. L5:1287 weakly suppresses
  19. L5:2029 weakly suppresses
  20. L2:2809 weakly excites
  21. L3:695 weakly suppresses
  22. L0:367 weakly suppresses
  23. L3:1808 weakly suppresses
  24. L4:659 weakly excites

Feeds later layers

  1. L11:2178 suppresses
  2. L8:2777 suppresses
  3. L9:118 excites
  4. L8:2298 suppresses
  5. L8:1944 suppresses
  6. L12:892 suppresses
  7. L9:1468 weakly suppresses "however" as a mid-sentence aside
  8. L8:142 weakly suppresses
  9. L13:1595 weakly excites
  10. L8:2707 weakly suppresses
  11. L8:2513 weakly suppresses
  12. L20:1821 weakly suppresses
  13. L14:396 weakly excites
  14. L22:2343 weakly excites
  15. L15:137 weakly suppresses
  16. L9:2714 weakly suppresses
  17. L19:2693 weakly suppresses
  18. L16:1923 weakly excites
  19. L11:2941 weakly suppresses
  20. L16:3000 weakly excites
  21. L13:742 weakly suppresses
  22. L9:2284 weakly suppresses
  23. L19:2359 weakly suppresses
  24. L16:2212 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.

2 values between -3.13 and -2.961 values between -2.80 and -2.641 values between -1.83 and -1.672 values between -1.50 and -1.342 values between -1.34 and -1.184 values between -1.18 and -1.026 values between -1.02 and -0.8611 values between -0.86 and -0.6910 values between -0.69 and -0.5330 values between -0.53 and -0.37119 values between -0.37 and -0.21590 values between -0.21 and -0.041999 values between -0.04 and 0.12844 values between 0.12 and 0.28201 values between 0.28 and 0.4444 values between 0.44 and 0.6117 values between 0.61 and 0.779 values between 0.77 and 0.931 values between 0.93 and 1.092 values between 1.09 and 1.254 values between 1.25 and 1.422 values between 1.42 and 1.582 values between 1.74 and 1.901 values between 2.06 and 2.231 values between 2.39 and 2.551 values between 2.55 and 2.711 values between 2.88 and 3.041 values between 3.20 and 3.36offnegativepositiveoften

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. typeid
  2. modo
  3. ervised
  4. ditor
  5. AdapterManager
  6. ��
  7. edition
  8. MessageType
  9. 亚
  10. ferences

Suppresses

  1. iz
  2. ht
  3. ak
  4. not
  5. OK
  6. 혁
  7. 北
  8. nut
  9. ACK
  10. fact