Neuron Explorer

Qwen3-0.6B

'to' after a noun (a role to play)

A neuron in layer 9 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 'to' when it comes right after a noun and introduces a task, like 'a role to play', 'a story to tell' or 'so much to gain'. When it was turned up, the replies did not clearly change in a visible way.

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

    this search on Wikipedia, Wiktionary, Google, OneLook.<|endoftext|>More than 3 million soldiers fought in the Civil War, and each had a story to tell. Some of those stories have been preserved through personal journals. We recently came

  2. #2 peak

    . The McKinsey study carried out for the ICCA on the contribution chemicals make to carbon reduction lays out much of its argument.⏎Chemistry has an important role to play in enabling sustainable solutions, is the conclusion of a book written by Robert K

  3. #3 peak

    were concluded in December 1993, multilateral trade negotiations were seen as a preserve of the developed countries and the developing countries have only a marginal role to play in the negotiation process; they were primarily the recipients of preferential market access

  4. #4 peak

    conflicts for the patient’s time to schedule and conduct therapy with any and all of these providers. It is important to recognize that every service provider has an important role to play and must be respectful of each other’s scheduling needs.⏎Basically, insurance companies

  5. #5 peak

    , motion and sound. A whole wide world of electronics and coding is waiting for you, and it fits in the palm of your hand.⏎Have an amazing project to share? The Electronics Show and Tell is every Wednesday at 7pm ET!

  6. #6 peak

    sstatement: said, “This is the inspiration for this year’s theme, Gender equality benefits everyone. It is a powerful reminder that we all have so much to gain when we advance gender equality. It is both the right thing to do and

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 is the number of times in 100 years? 100 years is 100 years. So, 100 years is 1 year. 1 year is 1 year. 1 year

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

    a 12-week, double-blind, randomized study – the most rigorous of clinical trial designs. After initial assessments, the patients received treatment with either clozapine or “high-dose” olanzapine (doses that exceed the

  2. #2 peak

    of this effort. We hope to have a massive impact on improving the tests clinicians have at hand to help them know whether their patients have TB, and what kind of treatment is most likely to benefit them.⏎The global TB research community, under the

  3. #3 peak

    ) and were monitored for improvement in their symptoms.⏎The researchers discovered that clozapine was approximately twice as likely to produce a treatment response as olanzapine.

  4. #4 peak

    of his own sin, Saul found himself in a hopeless predicament, and killed himself. Only in the sense that God’s word was the immovable object against which Saul rebelled could it be said that God “put him to death.” In

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. L5:1428 suppresses
  2. L7:2503 suppresses
  3. L7:1451 excites
  4. L6:2297 excites
  5. L6:958 suppresses
  6. L8:2276 suppresses
  7. L7:1634 suppresses
  8. L5:2633 excites
  9. L7:1974 weakly excites
  10. L8:876 weakly excites
  11. L5:663 weakly excites
  12. L6:1371 weakly suppresses
  13. L6:2271 weakly excites art, paintings and painters
  14. L7:1593 weakly suppresses
  15. L7:1849 weakly excites
  16. L7:978 weakly excites
  17. L7:1494 weakly suppresses
  18. L2:41 weakly suppresses
  19. L0:608 weakly excites
  20. L1:1897 weakly suppresses
  21. L7:2227 weakly excites
  22. L1:2121 weakly excites
  23. L6:2970 weakly suppresses
  24. L5:3014 weakly suppresses

Feeds later layers

  1. L15:2692 excites 'to' after a noun, before a purpose verb
  2. L10:1211 excites
  3. L15:1344 excites
  4. L13:938 excites
  5. L17:1755 excites
  6. L15:1851 excites 'that' after a noun, 'being' in passives
  7. L17:1012 excites
  8. L10:678 excites
  9. L21:2112 excites
  10. L14:2420 excites
  11. L19:2621 excites
  12. L18:1243 excites
  13. L20:1248 excites
  14. L13:1099 excites
  15. L13:46 weakly excites
  16. L21:692 weakly suppresses
  17. L18:1672 weakly excites
  18. L18:2988 weakly excites
  19. L19:2179 weakly excites
  20. L19:3052 weakly excites
  21. L14:2055 weakly excites preposition after a noun phrase
  22. L13:364 weakly suppresses
  23. L14:1452 weakly suppresses
  24. L20:94 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.

2 values between -1.05 and -0.942 values between -0.94 and -0.821 values between -0.82 and -0.711 values between -0.71 and -0.6015 values between -0.60 and -0.4921 values between -0.49 and -0.37109 values between -0.37 and -0.26360 values between -0.26 and -0.15993 values between -0.15 and -0.041440 values between -0.04 and 0.08641 values between 0.08 and 0.19228 values between 0.19 and 0.3054 values between 0.30 and 0.4117 values between 0.41 and 0.534 values between 0.53 and 0.642 values between 0.64 and 0.751 values between 0.75 and 0.863 values between 0.86 and 0.983 values between 0.98 and 1.092 values between 1.09 and 1.202 values between 1.20 and 1.311 values between 1.31 and 1.421 values between 1.42 and 1.541 values between 1.54 and 1.651 values between 1.87 and 1.991 values between 2.32 and 2.442 values between 3.33 and 3.45offnegativepositiveoften

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. 感兴趣的
  2. ffield
  3. omas
  4. consideration
  5. consulted
  6. ált
  7. noted
  8. upal
  9. ptune
  10. addock

Suppresses

  1. це
  2. 极端
  3. iese
  4. CE
  5. ners
  6. 各级
  7. aantal
  8. 中央
  9. 这种情况
  10. cest