Neuron Explorer

Qwen3-0.6B

'still' after 'but' in contrast

A neuron in layer 13 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 'still' when it follows a 'but' or 'although' clause, as in 'but still a huge problem'. It marks a point where something is true despite a contrast. Turning it up sometimes gave text with 'still' and a 'but' twist, though not consistently.

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

    3% higher than the death rate for white people. In 2016, it was 16% higher. An improvement, yes — but still a huge discrepancy.⏎As with everything related to cancer, the factors contributing to these

  2. #2 peak

    increased from 8% to 84%, and that the number of reported head injuries and potentially serious head injuries decreased.⏎The percentage differences were minor, but still noteworthy.⏎A similar analysis was conducted by the NHS in the United Kingdom. They

  3. #3 peak

    of the living room, and several measures were made on this area. Although increased insulation improved the noise dampening capabilities of the embedded pipelines, the noise was still heard. Also the improvements could not explain the fact that the noise was also heard

  4. #4 peak

    oscillator producing a sine wave of about 37 kHz, which is then boosted to about 2400 volts. That’s pretty good, but nowhere near the 15 kilovolts needed for a self

  5. #5 peak

    31, and then slide steeply again until early 1933 before rebounding almost vertically for a few months, and then at a slower but still rapid pace.⏎Also See | The Recession Alphabet (Graphics)⏎The deep

  6. #6 peak

    set to give us plenty more of.⏎As expected, the hibernating (or desiccated) tardigrades fared better, but they’re still far from immortal. Their median lethal temperature was 82.7° C (

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: What is the current, which is the, which is the, which is the, which is the, which is the, which is the, which is the, which is the, which is the, which is the, which is 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

    's disease, and how do you differentiate between regular memory loss and memory loss associated with Alzheimer's disease?⏎Answer: So, the memory changes of Alzheimer's disease represent a change for that person in

  2. #2 peak

    determine how newspaper articles and television reports describe and analyze the event. Use this...⏎6th - 12th English Language Arts CCSS: Adaptable⏎The State of the Union Address⏎Introduce your future voters to the annual State

  3. #3 peak

    such as product listings, were added. The new edition has also been changed by permitting the installation of decorative shrouds at the termination of a factory-built chimney<|endoftext|>When famous characters or people like Hercule Poirot

  4. #4 peak

    in which the student will answer two questions [50%].⏎Assessed Essay 1 will contribute 15% of the final mark for the module.<|endoftext|>Study the following vehicle parts for the type of vehicle you will be using during the

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. L7:2736 weakly suppresses
  2. L12:2622 weakly excites
  3. L9:2777 weakly suppresses
  4. L12:2770 weakly suppresses
  5. L7:2832 weakly suppresses
  6. L0:637 weakly suppresses
  7. L5:1080 weakly suppresses
  8. L3:1223 weakly excites
  9. L9:512 weakly suppresses
  10. L11:482 weakly excites
  11. L7:2018 weakly suppresses
  12. L9:962 weakly excites
  13. L11:2749 weakly excites
  14. L8:3020 weakly excites
  15. L5:1512 weakly excites
  16. L10:2371 weakly suppresses
  17. L6:2015 weakly excites
  18. L4:1663 weakly excites
  19. L7:809 weakly excites
  20. L7:126 weakly excites
  21. L2:2380 weakly suppresses
  22. L7:2738 weakly suppresses
  23. L5:2110 weakly excites
  24. L4:1183 weakly suppresses

Feeds later layers

  1. L14:1123 excites
  2. L16:2778 excites
  3. L14:1285 excites
  4. L14:1361 excites
  5. L14:2809 excites
  6. L14:2171 weakly suppresses
  7. L18:2340 weakly excites
  8. L14:536 weakly suppresses
  9. L21:864 weakly suppresses 'concentrated' and comparative adverbs
  10. L15:1422 weakly excites
  11. L14:2034 weakly excites
  12. L14:1903 weakly excites
  13. L14:2860 weakly suppresses
  14. L15:250 weakly excites
  15. L15:649 weakly excites
  16. L16:2748 weakly suppresses
  17. L14:1131 weakly suppresses
  18. L14:1057 weakly excites
  19. L23:628 weakly excites
  20. L20:1268 weakly suppresses
  21. L14:2712 weakly suppresses
  22. L19:1454 weakly excites
  23. L18:2883 weakly excites
  24. L15:2078 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 -1.63 and -1.531 values between -1.07 and -0.984 values between -0.98 and -0.896 values between -0.80 and -0.716 values between -0.71 and -0.6233 values between -0.62 and -0.5254 values between -0.52 and -0.43116 values between -0.43 and -0.34314 values between -0.34 and -0.25564 values between -0.25 and -0.16823 values between -0.16 and -0.06936 values between -0.06 and 0.03548 values between 0.03 and 0.12286 values between 0.12 and 0.21113 values between 0.21 and 0.3051 values between 0.30 and 0.4017 values between 0.40 and 0.498 values between 0.49 and 0.585 values between 0.58 and 0.675 values between 0.67 and 0.765 values between 0.76 and 0.863 values between 0.95 and 1.041 values between 1.04 and 1.133 values between 1.22 and 1.311 values between 1.31 and 1.411 values between 1.59 and 1.681 values between 1.68 and 1.772 values between 1.96 and 2.05offnegativepositiveoften

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. fires
  2. ernes
  3. že
  4. ILED
  5. ируем
  6. enes
  7. 层层
  8. enie
  9. anje
  10. grease

Suppresses

  1. 湛
  2. views
  3. .addAttribute
  4. 市教育局
  5. ToShow
  6. /o
  7. Admission
  8. !/
  9. chestra
  10. views