Neuron Explorer

Qwen3-0.6B

blank before a number in citations

A neuron in layer 5 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 gap just before a number in legal citations, dates and verse references, such as "§ 1395" or "March 3, 1879". Its negative side appears in hyphenated or underscored words. Turning it up gave unrelated text, so the steering does not confirm the pattern.

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 Section 1185 of Title 23 of the Vermont Statutes Annotated. And the federal citation 42 U.S.C. § 1395 tells us that this cited federal statute can be found in Title

  2. #2 peak

    be readily restored to shoot, automatically more than one shot, without manual reloading, by a single function of the trigger.” 26 U.S.C. § 5845(b). The machine gun was first widely used during World War

  3. #3 peak

    history, the U.S. Civil War, was resolved by amendments—creating the legal foundation for the equal rights of African Americans. The 13th, 14th and 15h Amendments—collectively known as the Civil

  4. #4 peak

    the Lord for his unfailing love and his wonderful deeds for mankind” (107:1, 2, 8, New International Version).⏎This is a recognition that God,

  5. #5 peak

    Irene and Tropical Storm Lee,<|endoftext|>Back to: Lesson 40⏎Go on to: Lesson 42⏎Key Points: Pieces 83 & 84⏎Pieces 83 is a great little tune but difficult. Pay

  6. #6 peak

    second director). On March 3, 1879, the 45th Congress and President Hayes agreed to

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) - but I can't, so? - This is a question about a question, but it's not just a question. (the) - This is a question about a question, but it's not just a question

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

    where_x, float where_y, float where_z);⏎void deallocate_model(model_3d_ref kill_this);⏎The compiler doesn't read your code like this -

  2. #2 peak

    Lab found healthy certified buildings pull in effective rents that are 4.4% to 7.7% more per square foot than non-certified and non-registered peer buildings.⏎What’s good for the planet and what’s good for people

  3. #3 peak

    The reason? Google and its ever-changing algorithms. Specifically, one of the latest actions Google had taken to fight fraudulent, insecure websites and pages is to label non-HTTPS pages as “not secure”:⏎There’s a high chance you might have

  4. #4 peak

    that are IC rated and those that are not IC rated. The difference between the two ratings is that the IC rated fixtures can be buried in insulation and the non-IC rated fixtures cannot be covered or in contact with insulation. The reason for this

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. L4:53 excites blank slot before a number
  2. L0:3025 weakly excites
  3. L4:1056 weakly excites
  4. L3:820 weakly excites
  5. L3:2152 weakly suppresses
  6. L2:3028 weakly excites
  7. L3:2759 weakly suppresses
  8. L4:859 weakly excites
  9. L4:71 weakly suppresses
  10. L4:2838 weakly excites
  11. L3:2009 weakly suppresses
  12. L3:858 weakly excites
  13. L2:944 weakly suppresses
  14. L0:2391 weakly excites
  15. L4:232 weakly suppresses
  16. L0:231 weakly suppresses
  17. L2:2192 weakly suppresses
  18. L4:2344 weakly suppresses
  19. L3:1018 weakly excites
  20. L2:543 weakly excites
  21. L2:424 weakly excites
  22. L4:1748 weakly suppresses
  23. L3:2111 weakly suppresses
  24. L2:716 weakly suppresses

Feeds later layers

  1. L14:140 excites
  2. L11:300 weakly suppresses
  3. L7:8 weakly suppresses
  4. L16:396 weakly excites
  5. L19:1351 weakly suppresses
  6. L19:786 weakly excites
  7. L18:2578 weakly excites
  8. L8:162 weakly suppresses
  9. L8:111 weakly excites the space before a year in dates
  10. L19:1254 weakly excites
  11. L13:2166 weakly suppresses
  12. L17:1161 weakly suppresses
  13. L12:613 weakly excites
  14. L24:197 weakly excites
  15. L6:152 weakly suppresses
  16. L26:1587 weakly excites
  17. L18:76 weakly excites blank spot before a number
  18. L26:840 weakly suppresses
  19. L20:59 weakly excites
  20. L22:164 weakly suppresses
  21. L6:2828 weakly suppresses
  22. L6:2 weakly suppresses
  23. L22:52 weakly excites
  24. L14:348 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 -2.45 and -2.351 values between -1.96 and -1.861 values between -1.57 and -1.474 values between -0.68 and -0.5810 values between -0.58 and -0.4814 values between -0.48 and -0.3938 values between -0.39 and -0.29101 values between -0.29 and -0.19373 values between -0.19 and -0.091702 values between -0.09 and 0.011233 values between 0.01 and 0.11309 values between 0.11 and 0.2181 values between 0.21 and 0.3019 values between 0.30 and 0.407 values between 0.40 and 0.502 values between 0.50 and 0.605 values between 0.60 and 0.701 values between 0.79 and 0.892 values between 0.89 and 0.991 values between 0.99 and 1.091 values between 1.19 and 1.291 values between 1.29 and 1.381 values between 1.38 and 1.48offnegativepositiveoften

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. ColumnType
  2. tures
  3. ortal
  4. geber
  5. agnar
  6. cctor
  7. onent
  8. FTER
  9. вещ
  10. 治理体系

Suppresses

  1. ec
  2. fe
  3. etest
  4. ้น
  5. ử
  6. Enlarge
  7. eload
  8. masks
  9. ep
  10. suppress