Neuron Explorer

Qwen3-0.6B

the word 'if'

A neuron in layer 22 of Qwen3-0.6B. Its strongest snippets are fairly consistent: several of them peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires strongly on the word 'if', especially when it starts a sentence or stands alone as a word. When it was turned down, the model got stuck repeating 'if if if', which fits the idea that it pushes against that word.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    Key words: tuna / Fish Aggregating Devices / behaviour / horizontal movements / ultrasonic telemetry / French Polynesia⏎© Elsevier, Inra, Ifremer, Cemagref, Ird, Cnrs,

  2. #2 peak

    James McIntyre⏎Read this poem in other languages⏎This poem has not been translated into any other language yet.⏎Still I Rise⏎The Road Not Taken⏎If You Forget Me⏎Edgar Allan Poe⏎Stopping By Woods On A Snowy

  3. #3 peak

    , New York. It helped to launch the U.S. campaign for women’s suffrage, or the right to vote.<|endoftext|>Have you ever wondered?⏎- If animals such as giraffes and

  4. #4 peak

    ‘raw’ in ‘straw’ or ‘if’ in ‘life’, as they are bound to appear double in the playing field

  5. #5 peak

    natural by-product of gas<|endoftext|>DataTableCollection.Item Property (String)⏎Gets the DataTable object with the specified name.⏎Assembly: System.Data (in System.Data.dll)⏎If a DataTable name exists that matches the case of the search string, it is returned

  6. #6 peak

    of Logical Knowledge. M.D.S. Braine, The "Natural Logic" Approach to Reasoning. E.K. Scholnick, The Three Faces of If. F.B. Murray, The

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: if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if if

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

    auditing, billing, and logging. In addition, the architecture also offers enterprises the flexibility of having agile business processes, better addresses the regulatory requirements like Sarbanes Oxley (SOX), and changes individual services without affecting other services.⏎The reality

  2. #2 peak

    part of the body in less than 72 hours. MRSA is strong - resistant to nearly all antibiotics, including penicillin, amoxicillin, oxacillin and methicillin. It causes more deaths in the U.S.

  3. #3 peak

    muscles, sensations, behaviour, emotions, consciousness, or a combination of these. The seizures can be prevented in most people by suitable antiepileptic medication. Oxcarbazepine works by stabilising the electrical activity of your brain, which

  4. #4 peak

    it is added to other white powders to increase dealers' margins and users' highs. While deaths due strictly to the abuse of prescription opioids, such as OxyContin, peaked around 2017, fentanyl-involved deaths have

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. L19:2348 excites
  2. L17:1282 excites
  3. L0:1470 excites
  4. L6:2134 excites
  5. L18:1617 excites
  6. L9:306 suppresses
  7. L20:466 weakly suppresses
  8. L17:619 weakly excites
  9. L19:1580 weakly excites
  10. L15:851 weakly excites
  11. L9:75 weakly excites 'what' or 'how' after see or understand
  12. L20:342 weakly suppresses
  13. L13:2966 weakly suppresses
  14. L1:1850 weakly suppresses percent-encoded url characters
  15. L0:921 weakly excites
  16. L15:740 weakly excites
  17. L20:2699 weakly excites
  18. L4:265 weakly suppresses
  19. L3:2440 weakly suppresses
  20. L11:502 weakly excites
  21. L8:132 weakly suppresses
  22. L18:1389 weakly excites
  23. L14:1324 weakly suppresses
  24. L16:1564 weakly excites

Feeds later layers

  1. L23:298 excites
  2. L26:509 suppresses
  3. L23:449 suppresses
  4. L27:511 weakly suppresses
  5. L25:768 weakly suppresses
  6. L26:1599 weakly suppresses
  7. L27:2280 weakly excites
  8. L24:1474 weakly suppresses
  9. L26:217 weakly suppresses
  10. L25:234 weakly excites
  11. L26:715 weakly excites
  12. L25:2302 weakly suppresses
  13. L24:123 weakly suppresses
  14. L23:1815 weakly suppresses
  15. L27:717 weakly excites
  16. L23:2750 weakly suppresses
  17. L24:1581 weakly suppresses
  18. L23:999 weakly suppresses "be" after modal verbs
  19. L26:466 weakly excites
  20. L27:2219 weakly excites
  21. L27:2209 weakly suppresses
  22. L24:2631 weakly suppresses
  23. L24:2423 weakly suppresses
  24. L23:1068 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 -8.30 and -6.661 values between -6.66 and -5.031 values between -5.03 and -3.4010 values between -3.40 and -1.771119 values between -1.77 and -0.142656 values between -0.14 and 1.5054 values between 1.50 and 3.1325 values between 3.13 and 4.769 values between 4.76 and 6.4010 values between 6.40 and 8.033 values between 8.03 and 9.662 values between 9.66 and 11.293 values between 11.29 and 12.932 values between 12.93 and 14.563 values between 21.09 and 22.721 values between 25.98 and 27.621 values between 27.62 and 29.251 values between 32.51 and 34.151 values between 37.41 and 39.042 values between 45.57 and 47.211 values between 47.21 and 48.841 values between 50.47 and 52.101 values between 55.37 and 57.00offnegativepositiveoften

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. EVER
  2. possible
  3. DebugEnabled
  4. {[
  5. orgot
  6. you
  7. POSS
  8. ighting
  9. 您
  10. 吒

Suppresses

  1. 如果
  2. 如果有
  3. 但如果
  4. ificantly
  5. 如果不是
  6. 如果我们
  7. 如果
  8. if
  9. ificate
  10. ivement