Neuron Explorer

Qwen3-0.6B

evaluative words: good, bad, positive

A neuron in layer 10 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 words that judge something as good or bad, such as 'favorable', 'good terms', 'bad rap' and 'positively'. Turning it up did not produce a clear change in what the model wrote about.

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

    is too jaded by his ambition.⏎Although this essay is historically accurate it lacks important details, which might paint a different view of Columbus. Boorstin writes favorable of Columbus and depicts him as a heroic and determined figure who helped shape history,

  2. #2 peak

    feuding families eventually stopped but only because Romeo and Juliet died. This was mainly because of the Friar.⏎The Capulets and Montagues were not on good terms with each other. Unfortunately though, Romeo and Juliet fell in love with each

  3. #3 peak

    age or time. Teach them that a good person chooses to think good thoughts, not to say profaned words, do good deeds, go to good places

  4. #4 peak

    of heart disease.⏎Carbohydrates have gotten a bad rap lately, influenced by various fad diets popular today. As a result,

  5. #5 peak

    the studies.⏎This makes one wonder if the collagen used in<|endoftext|>The passive voice in Spanish is one of those strange grammatical expressions that strikes fear into the best of us. The truth of the matter is that one of the challenges of learning a

  6. #6 peak

    , Irish is the expression of a cherished culture, but its close association with nationalism and republicanism means that Protestants and unionists rarely see it in a positive light. History reveals that this was not always the case. For centuries, Protestants

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: $ S_n = d_k h_k l_k h_k l_0 h_k l_0 h_k l_0 h_0 h_k l_0 h_k l_0 h_0 h_0 h_k l_

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

    was written amidst the 1800’s. Hawthorne was a famous American author during that time frame. He is a relative to a judge from the Salem Witch Trials, which was his Great-Great Grandfather John Hathorne. Hath

  2. #2 peak

    life forces and increase energy and strength. Ruby is also known as the stone of courage, and legends record that a person possessing a ruby can walk through life without fear, evil or experiencing misfortune.⏎Throughout gemstone history, spinel and garn

  3. #3 peak

    more than one anxiety disorder. Sometimes anxiety results from a medical condition that needs treatment.⏎Common symptoms of anxiety would be increased

  4. #4 peak

    over on a stool in the 15th Evacuation Hospital in Sicily. A husband and father from South Bend, Indiana, had fought bravely for eight months through the Tunisian and Sicilian campaigns. Within the last two weeks

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. L7:2333 excites
  2. L7:1110 suppresses
  3. L6:2241 excites
  4. L7:1674 suppresses
  5. L8:2784 excites
  6. L6:1821 suppresses
  7. L6:2847 suppresses
  8. L8:1978 suppresses
  9. L8:2637 suppresses
  10. L8:2279 excites
  11. L1:1263 suppresses
  12. L9:667 suppresses
  13. L7:1586 suppresses
  14. L7:1839 weakly excites
  15. L8:2899 weakly suppresses
  16. L8:1143 weakly suppresses
  17. L9:2963 weakly excites
  18. L6:222 weakly suppresses
  19. L7:1769 weakly suppresses
  20. L9:1623 weakly suppresses
  21. L9:258 weakly excites
  22. L9:2014 weakly suppresses
  23. L8:1071 weakly excites
  24. L4:1476 weakly suppresses

Feeds later layers

  1. L11:2178 excites
  2. L20:1384 excites
  3. L16:2617 excites
  4. L13:2350 suppresses
  5. L14:2623 excites
  6. L12:1749 excites
  7. L24:374 excites
  8. L19:2952 excites
  9. L18:2051 excites
  10. L13:1981 excites
  11. L15:2970 weakly excites
  12. L16:2756 weakly excites
  13. L12:2764 weakly excites
  14. L12:1886 weakly excites
  15. L23:1592 weakly excites
  16. L14:1286 weakly excites
  17. L21:1689 weakly excites
  18. L18:2187 weakly excites
  19. L21:2186 weakly excites
  20. L15:591 weakly suppresses
  21. L11:971 weakly excites
  22. L13:2414 weakly excites
  23. L11:2123 weakly excites
  24. L18:1698 weakly 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.

2 values between -1.31 and -1.152 values between -1.15 and -0.981 values between -0.98 and -0.823 values between -0.82 and -0.6618 values between -0.66 and -0.50190 values between -0.50 and -0.33935 values between -0.33 and -0.171621 values between -0.17 and -0.01848 values between -0.01 and 0.15144 values between 0.15 and 0.3261 values between 0.32 and 0.4824 values between 0.48 and 0.6419 values between 0.64 and 0.8010 values between 0.80 and 0.975 values between 0.97 and 1.138 values between 1.13 and 1.294 values between 1.29 and 1.452 values between 1.45 and 1.621 values between 1.62 and 1.781 values between 1.78 and 1.941 values between 1.94 and 2.104 values between 2.10 and 2.271 values between 2.27 and 2.431 values between 2.43 and 2.591 values between 2.59 and 2.751 values between 5.03 and 5.19offnegativepositiveoften

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. -negative
  2. 正面
  3. positive
  4. negative
  5. Negative
  6. 恶劣
  7. Positive
  8. adverse
  9. negative
  10. 负面

Suppresses

  1. elts
  2. @",
  3. \base
  4. taller
  5. 門
  6. @}
  7. byname
  8. recre
  9. bran
  10. towers