Neuron Explorer

Qwen3-0.6B

'or' in ranges and alternatives

A neuron in layer 21 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 reacts to the word 'or' when it joins two options or a range, such as 'age 14 or 15' or 'an hour or less'. Turning it up did not produce anything clearly about that, just wandering text.

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

    has found that most studies on college hunger reveal food insecurity rates among U.S. college students to be higher than the national average for adults not attending a two- or four-year institution.⏎When students are food-insecure, they are more likely to

  2. #2 peak

    Its greenhouse gas effect is huge with over twenty thousand times the effect of carbon dioxide over a hundred years and remaining in the atmosphere for a millennium or so.⏎In spite of these alarming figures its actual contribution at current levels of production

  3. #3 peak

    sensor on a piece of paper towel coated with water-resistant substance all over, except the areas showing four blood types and a positive or negative Rhesus factor. These areas are, packed with antibodies, interact with

  4. #4 peak

    this method will help you achieve a bouncing ball animation in probably 5 minutes or less.⏎So without wasting much time, let’s get into it!⏎Here are

  5. #5 peak

    Iowa. A lot of kids work for neighbors, and there are certain things parents would like their kids to be able to do when they reach age 14 or 15—including running a tractor and other farm machinery, Lang says.⏎The

  6. #6 peak

    MIT is one of the best universities in the world. Are you able to solve the problem that supposedly took an MIT professor a long time to solve in a day or less? Find out!⏎How can you spend your Math Minutes this week?⏎-

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: for me, the first time I will be 1000 times more often going to be: (1000) * (1000) * (1000) * (1000)

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

    While not providing enough of certain minerals is a concern, oversupplementation can also cause problems. Pet food manufacturers keep a close eye on just how much of each and every mineral is used in a dog food. Minerals used must also be able to

  2. #2 peak

    971 and 9.8 million in 2001, according to a report published in the March 11 issue of Morbidity and Mortality Weekly Report.⏎And the number of cancer survivors continues to grow because of

  3. #3 peak

    - 3 Pulse transformer⏎- 4 RF transformer⏎- 5 Audio transformer⏎- 6 Other types⏎- 7 See also⏎- 8 References⏎This is the most common type of transformer, widely used in electric power

  4. #4 peak

    keywords for each book and chapter.⏎If you think you should have access to this title, please contact your librarian.⏎To troubleshoot, please check our FAQs, and if you can't find the answer there, please contact us.<|endoftext|>Indigenous

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. L14:801 excites 'and' or 'or' joining list items
  2. L15:727 excites
  3. L12:51 excites
  4. L9:92 suppresses "and" or "or" joining two items
  5. L17:3025 suppresses
  6. L0:313 excites "and" joining paired nouns
  7. L6:427 suppresses
  8. L4:57 suppresses
  9. L5:1469 suppresses
  10. L4:108 excites "or" offering a choice in web text
  11. L18:1386 excites unclear: words before 'and' or 'or'
  12. L9:2010 suppresses
  13. L13:131 excites 'and' joining parallel phrases
  14. L3:250 excites
  15. L15:217 suppresses
  16. L8:70 excites 'and' between stacked superlatives
  17. L16:376 suppresses 'or' joining alternatives
  18. L9:77 excites commas in citations and name lists
  19. L12:1235 suppresses commas inside numbers and lists
  20. L18:2197 suppresses
  21. L7:999 suppresses
  22. L16:2940 excites commas inside lists and names
  23. L11:682 excites gap before a second number
  24. L1:1250 suppresses

Feeds later layers

  1. L25:1694 weakly excites 'and' joining paired names or fields
  2. L22:198 weakly excites
  3. L26:1007 weakly excites
  4. L26:339 weakly suppresses
  5. L24:2406 weakly suppresses
  6. L27:1163 weakly suppresses
  7. L27:753 weakly suppresses
  8. L27:909 weakly excites
  9. L26:1997 weakly suppresses
  10. L26:62 weakly excites
  11. L24:80 weakly suppresses
  12. L27:2267 weakly suppresses
  13. L24:1977 weakly suppresses
  14. L27:743 weakly suppresses
  15. L26:274 weakly suppresses
  16. L25:446 weakly excites
  17. L24:2151 weakly suppresses
  18. L27:1233 weakly suppresses
  19. L25:2770 weakly suppresses
  20. L23:1172 weakly excites
  21. L23:1445 weakly excites
  22. L24:1951 weakly suppresses
  23. L22:767 weakly suppresses
  24. L27:2632 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 -6.62 and -5.435 values between -4.23 and -3.0415 values between -3.04 and -1.85232 values between -1.85 and -0.652947 values between -0.65 and 0.54321 values between 0.54 and 1.73104 values between 1.73 and 2.9360 values between 2.93 and 4.1228 values between 4.12 and 5.3232 values between 5.32 and 6.5120 values between 6.51 and 7.7019 values between 7.70 and 8.9022 values between 8.90 and 10.0916 values between 10.09 and 11.2812 values between 11.28 and 12.487 values between 12.48 and 13.6710 values between 13.67 and 14.878 values between 14.87 and 16.068 values between 16.06 and 17.257 values between 17.25 and 18.455 values between 18.45 and 19.645 values between 19.64 and 20.831 values between 20.83 and 22.032 values between 22.03 and 23.225 values between 23.22 and 24.411 values between 24.41 and 25.611 values between 25.61 and 26.801 values between 26.80 and 28.001 values between 28.00 and 29.191 values between 30.38 and 31.581 values between 32.77 and 33.964 values between 33.96 and 35.162 values between 35.16 and 36.352 values between 36.35 and 37.541 values between 37.54 and 38.741 values between 39.93 and 41.13offnegativepositiveoften

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. そうで
  2. /or
  3. eler
  4. IOR
  5. тому
  6. �
  7. orra
  8. ivers
  9. =post
  10. oin

Suppresses

  1. ируем
  2. 第一步
  3. ニア
  4. ву
  5. 的第一个
  6. 第一位
  7. 第一天
  8. nx
  9. CompatActivity
  10. .'));⏎