Neuron Explorer

Qwen3-0.6B

'to' after 'so as' and infinitive cues

A neuron in layer 4 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' in phrases like 'so as to' or 'a party to', where a purpose or infinitive is about to follow. Turning it up gave no clear theme in the output.

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

    (cause to go off) "fire a gun"; "fire a bullet"⏎- S: (v) fire (bake in a kiln so as to harden) "fire pottery

  2. #2 peak

    educational computers are so busy trying to maintain their market share that they failed to see the logic of working to the same basic voltage in selecting their architecture, so as to be more compatible with one another. If they were compatible

  3. #3 peak

    get to normal subduing the demand. The government starts to deepen the pocket by increasing taxes, so as to benefit from the economic boom.⏎In other words, the economy is too complex to

  4. #4 peak

    the need for any investigation into the likelihood or suspicion of bias. The most common form of imputed bias is where the judge or the decision-maker is a party to a suit – in other words they are actually involved in the dispute that appears before

  5. #5 peak

    for show-up. He or she can be required to speak, stand in a particular pose, and wear appropriate clothing.⏎Suspects can invoke the Sixth Amendment right to counsel during lineups and showups held after they have been arraigned or indicted

  6. #6 peak

    capital Kt: with BK, BKK >0.⏎Consumers save in the form of capital; the representative consumer chooses an optimal consumption plan, so as to maximize⏎where c is consumption per capita and r is the rate of time preference

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: I have to get some sleep. I have to get to the point of sleep. It is not a thing to think that sleep, and to get to the point of sleep. I have to get to the point of sleep. It is

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

    there's great hope⏎If we our eyes do only ope.⏎With joy they will often glisten,⏎If to truth one doth but listen ;⏎But people often turn deaf ear⏎And what

  2. #2 peak

    that compensates for the impairment of functional networks caused by brain damage. A reserve that is constituted from childhood and that we can continue to feed throughout life.⏎And to stimulate the brain there is no mystery: You have to use it. Those who

  3. #3 peak

    was disclosed.⏎She let the cat out of her bag of verse she almost proposed to her hero in rhyme.George Meredith: The Egotist, iii.⏎To bell the cat. (See page 119, B

  4. #4 peak

    set forth in the “preamble,” or opening clause, of the Constitution of the United States;⏎1. “To form a more perfect union;”⏎2. “To establish justice;”⏎3

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. L3:1152 suppresses
  2. L2:750 excites
  3. L3:1859 suppresses
  4. L1:906 suppresses
  5. L3:2705 suppresses
  6. L3:2947 suppresses
  7. L3:59 excites
  8. L3:1573 suppresses
  9. L1:2296 excites
  10. L3:1016 weakly excites
  11. L0:1551 weakly suppresses
  12. L2:1036 weakly excites
  13. L3:691 weakly excites
  14. L3:2818 weakly excites
  15. L3:456 weakly suppresses
  16. L2:953 weakly suppresses
  17. L3:50 weakly excites
  18. L0:1301 weakly suppresses
  19. L0:178 weakly suppresses
  20. L3:115 weakly suppresses
  21. L2:1899 weakly excites
  22. L2:126 weakly excites
  23. L3:311 weakly suppresses
  24. L2:2866 weakly suppresses

Feeds later layers

  1. L19:858 strongly suppresses
  2. L5:2251 suppresses
  3. L8:1627 suppresses
  4. L18:1001 suppresses
  5. L16:1168 suppresses 'to' after verbs of change
  6. L10:132 suppresses
  7. L9:723 suppresses
  8. L5:2538 suppresses
  9. L9:1890 suppresses
  10. L17:1755 suppresses
  11. L17:1297 suppresses
  12. L9:2267 suppresses 'with' after exchange nouns
  13. L12:2084 suppresses
  14. L11:437 suppresses
  15. L5:2026 suppresses "to" after made or need-type verbs
  16. L27:2960 suppresses
  17. L19:2395 suppresses
  18. L6:2686 suppresses
  19. L6:2306 weakly suppresses
  20. L6:2899 weakly suppresses
  21. L5:363 weakly excites
  22. L7:1014 weakly suppresses
  23. L17:1843 weakly suppresses
  24. L14:2972 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 -7.48 and -7.251 values between -6.58 and -6.351 values between -5.45 and -5.222 values between -5.22 and -5.001 values between -4.77 and -4.552 values between -4.32 and -4.101 values between -4.10 and -3.871 values between -3.65 and -3.422 values between -3.42 and -3.204 values between -3.20 and -2.974 values between -2.97 and -2.741 values between -2.74 and -2.522 values between -2.52 and -2.294 values between -2.29 and -2.075 values between -2.07 and -1.842 values between -1.84 and -1.625 values between -1.62 and -1.396 values between -1.39 and -1.179 values between -1.17 and -0.946 values between -0.94 and -0.727 values between -0.72 and -0.498 values between -0.49 and -0.2713 values between -0.27 and -0.042607 values between -0.04 and 0.191093 values between 0.19 and 0.41115 values between 0.41 and 0.642 values between 0.64 and 0.861 values between 0.86 and 1.091 values between 1.09 and 1.311 values between 1.31 and 1.54offnegativepositiveoften

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. bí
  2. ifter
  3. ou
  4. istinguished
  5. pip
  6. [('
  7. \$
  8. (\$
  9. ited
  10. Petr

Suppresses

  1. 阗
  2. hn
  3. Neighbors
  4. gether
  5. (rawValue
  6. ffee
  7. MMI
  8. )object
  9. 澍
  10. ollo