Neuron Explorer

Qwen3-0.6B

unclear: mixed mid-sentence words

A neuron in layer 17 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron responds to a wide mix of words in the middle of sentences, such as 'which', 'processors', 'already' and 'complex', with no obvious shared meaning. Turning it up gave repetitive or odd text, so it is hard to say what it does.

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

    4004 and i4040, Intellec 8 and Intellec 8 mod 80 for the 8-Bit processors.⏎Due to not calling the DEC-PDP-Computer "computer", the Int

  2. #2 peak

    an older man, seeing each day and the wonders of the natural world as gifts for which we should be grateful. With this video in mind, our teachers show their students

  3. #3 peak

    cavalry commanded by Lord Cardigan, and to recapture the guns. Lord Lucan and Lord Cardigan saw only the heavy guns in their front—those to which Lord Raglan referred being concealed from their view by the high ground. They,

  4. #4 peak

    the effects of Word Generation, a middle-school vocabulary intervention, on the learning, maintenance, and consolidation of academic vocabulary for students from English-speaking homes, proficient English speakers from language-minority homes, and limited English-proficiency students. Using individual

  5. #5 peak

    It’s the first time in recorded history that two strong tropical cyclones have hit Mozambique in the same season.⏎The communities impacted by Idai and Kenneth were already vulnerable to slow-onset climate emergencies, such as droughts, rising sea levels

  6. #6 peak

    two perpendicular components namely, IRMS cos φ along VRMS and IRMS sin φ perpendicular to VRMS as shown in Figure 4.56.⏎

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: A) All and $A \rightarrow \{1/2 a_0\+\a_0\ +\b+\2\c_2(2\c_0(2c_0(x=0\+

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

    down to horizontal. Sexes and ages are alike.Jackdaws undergo a complete moult from June to September in the western parts of their range, and a month later in the east. Immature birds are duller in plumage.

  2. #2 peak

    capillary tubes or fixed metering devices.⏎Liquid is then pumped out of the receiver into the liquid line. When the liquid under high pressure reaches the metering device, the cycle starts over.⏎While servicing

  3. #3 peak

    Write’ function.⏎Do RGB LEDs need PWM?⏎an RGB LED is a combination of three LEDs in just one package: red, green and blue; there are two kinds of RGB LEDs: common cathode and common anode RGB LEDs; you

  4. #4 peak

    ened and comfortable with initiating and dealing with resistance; it is a mark of true progress.⏎Take a look at what your school is focusing on now and ask these questions: Are we nibbling at the edges or are we considering some real game chang

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. L2:49 excites
  2. L6:13 suppresses
  3. L6:16 suppresses
  4. L3:115 excites
  5. L0:1224 excites
  6. L1:1464 suppresses
  7. L2:1710 excites
  8. L0:1301 excites
  9. L2:75 excites
  10. L8:7 suppresses
  11. L6:17 excites
  12. L3:59 suppresses
  13. L5:46 suppresses
  14. L0:178 excites
  15. L5:110 excites
  16. L6:12 suppresses
  17. L6:7 excites
  18. L8:16 excites
  19. L8:14 suppresses
  20. L1:2801 suppresses
  21. L7:122 suppresses
  22. L2:126 suppresses
  23. L1:2683 suppresses
  24. L1:460 suppresses

Feeds later layers

  1. L18:624 strongly suppresses
  2. L18:1488 strongly excites
  3. L18:823 strongly excites
  4. L18:1596 strongly suppresses
  5. L18:962 suppresses
  6. L18:247 excites
  7. L20:707 excites
  8. L18:2064 excites
  9. L20:677 excites
  10. L18:755 excites
  11. L22:245 suppresses
  12. L19:1773 excites commas in non-English text
  13. L18:1703 suppresses
  14. L19:627 suppresses
  15. L19:1209 excites unclear: commas after dates and clauses
  16. L18:1764 suppresses
  17. L18:1956 suppresses
  18. L21:966 suppresses
  19. L21:562 suppresses
  20. L19:788 excites
  21. L19:767 excites
  22. L19:2283 excites
  23. L18:2271 excites
  24. L20:1159 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 -8.25 and -7.882 values between -7.88 and -7.521 values between -7.52 and -7.155 values between -6.78 and -6.426 values between -6.42 and -6.058 values between -6.05 and -5.683 values between -5.68 and -5.3212 values between -5.32 and -4.9525 values between -4.95 and -4.5833 values between -4.58 and -4.2149 values between -4.21 and -3.8553 values between -3.85 and -3.4876 values between -3.48 and -3.11113 values between -3.11 and -2.75154 values between -2.75 and -2.38194 values between -2.38 and -2.01250 values between -2.01 and -1.65285 values between -1.65 and -1.28282 values between -1.28 and -0.91312 values between -0.91 and -0.55320 values between -0.55 and -0.18345 values between -0.18 and 0.19312 values between 0.19 and 0.56227 values between 0.56 and 0.92197 values between 0.92 and 1.29193 values between 1.29 and 1.66125 values between 1.66 and 2.02101 values between 2.02 and 2.3979 values between 2.39 and 2.7649 values between 2.76 and 3.1241 values between 3.12 and 3.4915 values between 3.49 and 3.8626 values between 3.86 and 4.223 values between 4.22 and 4.594 values between 4.59 and 4.963 values between 4.96 and 5.332 values between 5.33 and 5.691 values between 5.69 and 6.061 values between 6.06 and 6.43offnegativepositiveoften

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. EFR
  2. personals
  3. licts
  4. licht
  5. 脲
  6. 两个人
  7. categoryId
  8. ) ⏎
  9. iphery
  10. 钋

Suppresses

  1. a
  2. A
  3. ang
  4. N
  5. N
  6. A
  7. a
  8. n
  9. cal
  10. n