Neuron Explorer

Qwen3-0.6B

opening bracket before an acronym

A neuron in layer 25 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 lights up on the open bracket that comes right after a full name, just before its short form, like the "(" in "Test of Economic Literacy (TEL)". Turning it up made the writing wander into repeated, broken phrases, so the steering tells us little about it.

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

    ologic formations has been developed by Dr. von Wegner, a world-renowned senior geochemist at a national laboratory. The method, Thermal Oxygenation Xenolith (

  2. #2 peak

    a unique opportunity to make frequent assessments of economic literacy in U.S. classrooms. In this survey, student responses to test items from the Test of Economic Literacy (TEL) are used to create economic concept and content area achievement benchmarks. These benchmarks

  3. #3 peak

    Propulsion Laboratory manages the 2001 Mars Odyssey mission for NASA's Office of Space Science, Washington, D.C. The Thermal Emission Imaging System (THEMIS) was developed by Arizona State University, Tempe, in collaboration

  4. #4 peak

    Wahi) your nerves could not have endured it. Therefore, an interval was given in order to afford you peace and tranquillity." This state was experien

  5. #5 peak

    trillion in the current fiscal year that ends next September or nearly 5% of gross domestic product.⏎It’ll be the first time the deficit has topped the $1 trillion mark since 2012. Deficits topped $1 trillion for

  6. #6 peak

    to support WPA.⏎The WPA protocol implements much of the IEEE 802.11i standard. Specifically, the Temporal Key Integrity Protocol (TKIP) was

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 an in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in

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

    ia, which might be more disabling that the tics. Even though this complication has not been described in persons with Tourette's syndrome, the use of haloperidol is limited to the most serious cases.⏎Other medications in this group such

  2. #2 peak

    video:<|endoftext|>Tourette's Syndrome (cont.)⏎IN THIS ARTICLE⏎Treatment of Tics⏎The most effective medication for the suppression of tics is haloperidol (Haldol), a

  3. #3 peak

    and comet Kohoutek, whose sweep around the sun at that time was a scientific bonus. Daily

  4. #4 peak

    pre-exposure prophylaxis (PrEP) did not experience any protection against HIV compared to those in the placebo arm

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. L24:2959 strongly suppresses
  2. L24:2012 excites
  3. L23:245 suppresses
  4. L23:363 suppresses
  5. L23:2818 weakly suppresses
  6. L23:801 weakly excites
  7. L22:3005 weakly excites
  8. L7:1247 weakly excites
  9. L24:2858 weakly suppresses
  10. L14:107 weakly excites
  11. L16:2977 weakly excites 'which' after a preposition
  12. L22:234 weakly excites
  13. L10:1572 weakly excites
  14. L1:257 weakly suppresses
  15. L11:395 weakly suppresses
  16. L6:1819 weakly suppresses
  17. L13:2505 weakly excites
  18. L18:2421 weakly suppresses
  19. L16:1613 weakly excites
  20. L17:1173 weakly suppresses
  21. L8:2816 weakly excites
  22. L15:1049 weakly excites
  23. L4:1279 weakly suppresses
  24. L1:2713 weakly excites

Feeds later layers

  1. L26:977 suppresses
  2. L26:367 excites
  3. L27:844 weakly excites
  4. L26:1785 weakly excites
  5. L27:2457 weakly excites
  6. L27:686 weakly suppresses
  7. L27:2003 weakly suppresses
  8. L27:2303 weakly suppresses
  9. L27:2265 weakly suppresses
  10. L27:1571 weakly suppresses
  11. L27:420 weakly suppresses
  12. L27:174 weakly suppresses
  13. L27:611 weakly suppresses
  14. L27:2238 weakly suppresses
  15. L26:1005 weakly suppresses
  16. L26:2439 weakly suppresses
  17. L27:2221 weakly suppresses
  18. L27:1347 weakly suppresses
  19. L26:304 weakly excites
  20. L27:440 weakly suppresses
  21. L26:1789 weakly suppresses
  22. L27:2107 weakly suppresses
  23. L26:1474 weakly suppresses
  24. L27:934 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 -15.47 and -14.591 values between -13.72 and -12.851 values between -11.97 and -11.101 values between -11.10 and -10.221 values between -9.35 and -8.472 values between -6.72 and -5.853 values between -5.85 and -4.972 values between -4.97 and -4.106 values between -4.10 and -3.2212 values between -3.22 and -2.3521 values between -2.35 and -1.48109 values between -1.48 and -0.603322 values between -0.60 and 0.27374 values between 0.27 and 1.1517 values between 1.15 and 2.0215 values between 2.02 and 2.905 values between 2.90 and 3.772 values between 3.77 and 4.653 values between 4.65 and 5.522 values between 6.40 and 7.272 values between 7.27 and 8.152 values between 8.15 and 9.022 values between 10.77 and 11.641 values between 18.64 and 19.52offnegativepositiveoften

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. trap
  2. tra
  3. Tracy
  4. Tr
  5. Tristan
  6. Tr
  7. tf
  8. .tim
  9. Trades
  10. Travis

Suppresses

  1. APTER
  2. ィ
  3. umph
  4. activated
  5. 운
  6. 颔
  7. yyyy
  8. -faced
  9. cks
  10. activation