Neuron Explorer

Qwen3-0.6B

'or' in lists of alternatives

A neuron in layer 11 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 'or' when it offers a choice between things, such as 'either the arm or groin' or 'a friend, family member, or teacher'. Turning it up did not produce any clearly related writing.

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

    finds that it is in the best interests of the minor child(ren) in the case, the Judge can appoint a “Guardian ad Litem”. Either party to the case can request that the Judge appoint a Guardian ad Litem, or

  2. #2 peak

    determine that num is not divisible by 5 and 7, then increase⏎count by one.⏎Share your code⏎Show a friend, family member, or teacher what you've done!<|endoftext|>cigarette butts are the most common

  3. #3 peak

    an alpha of 10%. Show your hypotheses, calculation of F* using Excel, and either the critical F value(s) or p-value, calculated by Excel, and decision. Describe the difference if one exists

  4. #4 peak

    they enjoy mimicking the pronunciation of new and strange words; and they play with rhyming words through repetition eve.⏎Summary õ PDF, DOC, TXT or eBook ´ James J. AsherNt very clear and comes with ready

  5. #5 peak

    .⏎People who inherit the transmuted hemoglobin gene from both biological parents have sickle cell anemia. People who inherit the transmuted gene from one parent have the sickle cell trait.⏎What Raises One’s Risk for Sickle Cell

  6. #6 peak

    style and structure for their own alphabet books and choose a word for each letter from content area textbooks, encyclopedias, reference books, or suggested websites. A storyboard is constructed including each of the 26 words,

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 do I think... I have a lot, and I want to know... What is 100,000, 100,000, 100,000,

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

    1000). It stands for the number of expressions to follow. Each expression consists of a single line containing a positive integer number, an operator (one of +, - and *) and the second positive integer number. Every number has at

  2. #2 peak

    Based on your location, we recommend that you select: .⏎You can also select a web site from the following list:⏎Select the China site (in Chinese or English) for best site performance. Other

  3. #3 peak

    see local events and offers. Based on your location, we recommend that you select: .⏎You can also select a web site from the following list:⏎Select the China site (in Chinese or English) for best site performance. Other

  4. #4 peak

    uncertain? "Fluctuations in labor input in a RBC model, not caused by real wage fluctuations, should be negatively correlated with consumption."⏎Choose 1 out of the following 2. Provide a strict formal argument. Worth 25

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. L8:1475 suppresses
  2. L10:135 excites 'or' in alternative requirements
  3. L7:1475 excites 'or' joining alternatives in lists
  4. L10:901 suppresses
  5. L7:959 excites
  6. L8:2002 suppresses
  7. L10:1053 suppresses
  8. L4:108 excites "or" offering a choice in web text
  9. L10:1669 suppresses
  10. L9:2445 excites
  11. L7:2657 excites
  12. L4:57 excites
  13. L8:1220 excites
  14. L5:1505 excites
  15. L6:2644 weakly suppresses
  16. L8:519 weakly excites
  17. L6:2706 weakly suppresses
  18. L10:2583 weakly suppresses
  19. L7:2342 weakly suppresses
  20. L5:1744 weakly excites
  21. L10:1190 weakly excites
  22. L9:1066 weakly excites
  23. L9:2327 weakly excites
  24. L9:352 weakly suppresses

Feeds later layers

  1. L15:2935 excites
  2. L13:1183 suppresses
  3. L12:2217 excites
  4. L12:1278 suppresses
  5. L13:1387 excites
  6. L16:1718 suppresses
  7. L13:2226 excites
  8. L16:2950 suppresses
  9. L13:1015 excites
  10. L12:642 suppresses
  11. L13:1675 excites
  12. L13:1090 suppresses
  13. L13:553 excites
  14. L13:2807 excites
  15. L13:562 excites
  16. L14:2159 excites
  17. L18:1473 suppresses
  18. L14:1750 suppresses
  19. L14:2314 excites
  20. L14:1008 suppresses
  21. L14:1869 excites
  22. L18:2896 excites
  23. L15:247 weakly excites
  24. L12:1819 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.

1 values between -1.83 and -1.661 values between -1.66 and -1.503 values between -1.50 and -1.333 values between -1.33 and -1.171 values between -1.17 and -1.004 values between -1.00 and -0.837 values between -0.83 and -0.6748 values between -0.67 and -0.50145 values between -0.50 and -0.34627 values between -0.34 and -0.171609 values between -0.17 and -0.011236 values between -0.01 and 0.16147 values between 0.16 and 0.3225 values between 0.32 and 0.4913 values between 0.49 and 0.6611 values between 0.66 and 0.827 values between 0.82 and 0.991 values between 0.99 and 1.153 values between 1.15 and 1.322 values between 1.32 and 1.481 values between 1.65 and 1.811 values between 1.81 and 1.981 values between 1.98 and 2.152 values between 2.15 and 2.312 values between 2.31 and 2.482 values between 2.48 and 2.641 values between 2.64 and 2.811 values between 2.81 and 2.971 values between 3.14 and 3.301 values between 3.47 and 3.631 values between 4.63 and 4.79offnegativepositiveoften

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. nào
  2. 哪种
  3. lon
  4. omatic
  5. cdot
  6. singled
  7. qualche
  8. اث
  9. onal
  10. akah

Suppresses

  1. verständlich
  2. 两级
  3. 这两个
  4. 六个
  5. ово
  6. gänge
  7. 两个
  8. арам
  9. 一样
  10. avigation