Neuron Explorer

Qwen3-0.6B

the word a before long

A neuron in layer 25 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 word "a" when it is followed by "long" or "while", as in "for a long time". The logit lens points the same way, toward the word "long". Turning it up made the model repeat "long" over and over.

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

    is easy to find, looks beautiful in astro landscapes, and it contains a fantastic assortment of objects to photograph.⏎If you’ve been reading MilkyWayPhotographers for a while, you might think that the previous paragraph sounds a little familiar – well,

  2. #2 peak

    sharp drop in the number of Purple Martins in Ontario and the reason behind this is not clear. Figuring out where they go to escape our cold winters will go a long way in helping us understand why their numbers are dropping. To start answering

  3. #3 peak

    Aluminium is more affordable regarding prize. The advantage of stainless blades is that they are more durable than aluminum therefore once you buy them they are likely to serve you for a very long duration.⏎Various things should be put into consideration before purchasing a propeller

  4. #4 peak

    , but the animals are sometimes abused by careless owners or harsh conditions.⏎- While the use of animals in entertainment, largely against their will, has been around for a long time, there are more and more anti-cruelty laws protecting animals

  5. #5 peak

    _of_algorithms)⏎> 1 get a positive integer from input⏎> 2 if n> 10⏎> 3 print "This might take a while..."⏎> 4 for i = 1 to n⏎> 5

  6. #6 peak

    Gear, talks about sophisticated gear manufacturing methods and how they can help solve common gear-related problems.⏎Noisy gear trains have been a common problem for gear designers for a long time. With the demands for smaller gear boxes transmitting more power at higher rp

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: how can I not make me more...? I am not sure where to start. Let me go to the map... meosh, meosh, meosh. The result is that I have not made me more... meosh. The

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

    still in development, the technology has far-reaching implications in both aeronautics and robotics. Cory was also named Boeing Engineering Student of the Year yesterday at Hamphsire, England's Farnborough Airshow. Click here to read an

  2. #2 peak

    . Children who have fibromyalgia may also have pain or cramping in the lower abdomen with constipation

  3. #3 peak

    of antihistamines include drowsiness which is compounded in people with fibromyalgia. Decongestants are stimulants which the doctor says can

  4. #4 peak

    children and adolescents. Estimates suggest that juvenile-onset fibromyalgia affects 2% to 6% of schoolchildren, mostly adolescent

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. L21:2777 excites
  2. L22:1046 excites
  3. L23:143 suppresses
  4. L23:3034 excites
  5. L23:1982 weakly excites
  6. L3:1545 weakly suppresses
  7. L3:2356 weakly excites
  8. L18:2374 weakly suppresses
  9. L12:1880 weakly excites
  10. L10:13 weakly excites
  11. L23:380 weakly suppresses
  12. L13:504 weakly suppresses
  13. L22:1027 weakly suppresses
  14. L16:838 weakly excites
  15. L10:1385 weakly suppresses
  16. L24:712 weakly suppresses 'long' in 'long-term' and 'long-lasting'
  17. L5:1072 weakly suppresses
  18. L19:1712 weakly excites
  19. L3:494 weakly suppresses
  20. L16:1802 weakly excites
  21. L13:2687 weakly excites
  22. L15:2715 weakly suppresses
  23. L3:249 weakly excites
  24. L1:1467 weakly excites

Feeds later layers

  1. L27:1178 excites
  2. L26:1469 weakly excites
  3. L26:887 weakly suppresses
  4. L26:892 weakly suppresses
  5. L27:1830 weakly suppresses
  6. L26:2030 weakly suppresses
  7. L26:2721 weakly suppresses
  8. L26:2451 weakly suppresses
  9. L26:3002 weakly suppresses
  10. L26:1930 weakly suppresses
  11. L27:1583 weakly excites
  12. L27:797 weakly excites
  13. L26:2546 weakly suppresses
  14. L26:3007 weakly excites
  15. L26:2606 weakly suppresses
  16. L27:2653 weakly suppresses
  17. L27:1733 weakly excites
  18. L26:2337 weakly suppresses
  19. L26:568 weakly suppresses
  20. L26:2060 weakly excites
  21. L26:1674 weakly excites
  22. L27:1590 weakly suppresses
  23. L26:1655 weakly suppresses
  24. L26:2836 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 -26.84 and -25.131 values between -13.14 and -11.431 values between -11.43 and -9.722 values between -8.01 and -6.295 values between -6.29 and -4.589 values between -4.58 and -2.8741 values between -2.87 and -1.163601 values between -1.16 and 0.56192 values between 0.56 and 2.2721 values between 2.27 and 3.9812 values between 3.98 and 5.6910 values between 5.69 and 7.413 values between 7.41 and 9.122 values between 9.12 and 10.833 values between 10.83 and 12.541 values between 12.54 and 14.261 values between 15.97 and 17.681 values between 17.68 and 19.391 values between 39.94 and 41.66offnegativepositiveoften

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. 长
  2. long
  3. Long
  4. Long
  5. 長
  6. .long
  7. long
  8. _long
  9. (long
  10. →long

Suppresses

  1. 飨
  2. ender
  3. 匀
  4. xED
  5. бур
  6. Jaguar
  7. ус
  8. ser
  9. urus
  10. +)\