Neuron Explorer

Qwen3-0.6B

"in" or "real" after a verb

A neuron in layer 6 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 small connecting words like "in" right after an action word ("fly in tighter flocks", "operating in high temperatures") and on "real" in "real time". When it is turned up, the model's writing drifts in unrelated directions, so the effect is hard to see.

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

    and deeper wingbeats and proportionately narrower and less fingered wings. They also have a shorter, thicker neck, a much shorter bill and frequently fly in tighter flocks. The underwing is uniformly grey, unlike choughs.

  2. #2 peak

    turbomachinery is limited when operating in high temperature environments, which restricts the efficiency of gas turbines and internal combustion engines

  3. #3 peak

    line.)⏎- Sends alerts to drivers, staff and other stakeholders if someone is stuck in the gap.⏎- Uses existing camera infrastructure.⏎- Processes in real time, with low latency on a computer the size of a credit card (R

  4. #4 peak

    cannot provide full torque continuously when the rotor is stopped. This aids in maintaining the current position.⏎- Precise open-loop positioning and repetition. Stepper motors move in discrete steps as long as the motor stays under the maximum torque and current limits.

  5. #5 peak

    sense of responsibility towards them.⏎The companions<|endoftext|>The Internet of Things (IoT) is about connecting any device to the internet and enabling it to transmit data in real time without human intervention. This technology makes it possible for users to control their smart

  6. #6 peak

    "tuneful, beatful, and artful." This is someone who: is able to sing in tune as well as catch tunes and is comfortable singing alone and with others; is

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: the person who I hold up to, the man who I respect for, is not only the great, but also the person who I have been able to be. I have admired, and I have learned from, and I have respected,

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

    were real people" (11). He continues "it's plausible but unproven" and that is good enough for me.⏎Why do I love and admire Achilles so much? Let me count the ways. The "Greatest of the

  2. #2 peak

    all great heroes of action was the Greek Hercules, who might almost be regarded as idolised toil. The adoration of this semi-god actually exceeded the reverence accorded to many of the twelve gods of the Greek and Roman Olympus. This

  3. #3 peak

    acclaimed legend in her lifetime, a towering legacy left behind. Rest in peace, brave comrade,” she penned.⏎Former

  4. #4 peak

    God is the only thing worthy of worship. Any other gods that others believe in, then we would regard them as false

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. L2:1148 weakly suppresses
  2. L3:258 weakly suppresses
  3. L2:2032 weakly suppresses
  4. L5:2682 weakly excites
  5. L3:93 weakly suppresses
  6. L0:2762 weakly excites
  7. L4:1178 weakly excites
  8. L5:1936 weakly suppresses
  9. L4:1631 weakly suppresses
  10. L2:2250 weakly suppresses
  11. L0:2680 weakly suppresses
  12. L4:1059 weakly excites
  13. L4:1912 weakly excites
  14. L1:747 weakly suppresses
  15. L5:2334 weakly suppresses
  16. L0:2839 weakly suppresses
  17. L4:2881 weakly suppresses
  18. L2:2641 weakly excites
  19. L5:3010 weakly excites
  20. L4:1637 weakly suppresses
  21. L2:1166 weakly suppresses
  22. L2:1073 weakly suppresses
  23. L3:1017 weakly suppresses
  24. L4:2105 weakly suppresses

Feeds later layers

  1. L16:2658 weakly excites
  2. L7:2701 weakly excites
  3. L12:1601 weakly suppresses
  4. L7:1526 weakly excites
  5. L8:2225 weakly suppresses
  6. L9:2571 weakly suppresses
  7. L12:2764 weakly excites
  8. L17:1874 weakly excites
  9. L25:2219 weakly excites
  10. L23:2595 weakly suppresses
  11. L9:2723 weakly excites
  12. L18:1837 weakly excites
  13. L18:710 weakly suppresses
  14. L17:2395 weakly excites
  15. L25:635 weakly suppresses
  16. L22:574 weakly excites
  17. L18:2034 weakly excites
  18. L21:2482 weakly suppresses
  19. L19:1321 weakly suppresses
  20. L8:2661 weakly excites
  21. L18:2541 weakly suppresses
  22. L24:1233 weakly excites
  23. L13:2997 weakly excites
  24. L10:294 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 -5.19 and -5.021 values between -4.19 and -4.021 values between -3.03 and -2.861 values between -2.69 and -2.531 values between -2.36 and -2.201 values between -2.20 and -2.031 values between -2.03 and -1.861 values between -1.86 and -1.701 values between -1.70 and -1.532 values between -1.53 and -1.373 values between -1.20 and -1.033 values between -1.03 and -0.874 values between -0.87 and -0.709 values between -0.70 and -0.5323 values between -0.53 and -0.37194 values between -0.37 and -0.201002 values between -0.20 and -0.042041 values between -0.04 and 0.13531 values between 0.13 and 0.3068 values between 0.30 and 0.4612 values between 0.46 and 0.632 values between 0.63 and 0.802 values between 0.80 and 0.961 values between 0.96 and 1.132 values between 1.30 and 1.46offnegativepositiveoften

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. WXYZ
  2. OOK
  3. iday
  4. unny
  5. rary
  6. IJ
  7. ijk
  8. óng
  9. 衽
  10. proof

Suppresses

  1. امة
  2. ๆ
  3. 殿堂
  4. че
  5. foremost
  6. ellar
  7. Installed
  8. SqlCommand
  9. philosoph
  10. utzer