Neuron Explorer

Qwen3-0.6B

the word 'same' as a referring word

A neuron in layer 14 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 word "same" when it points back to something already mentioned, as in "the same principle" or "at the same time". When it was turned up, the model's writing drifted into loose, repetitive phrasing, which doesn't clearly show what it does.

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

    an "insider." Some of the papers discuss the issue of objectivity, with the implication that objectivity is best obtained by standing outside the religion. This same subtext continues throughout the papers: that there is something amiss with being a

  2. #2 peak

    agency within the U.S. Department of Labor, workplace fatalities have been cut in half and occupational injury and illness rates have declined 40 percent. At the same time, U.S. employment has nearly doubled.⏎Under the OSH Act,

  3. #3 peak

    2009, 18- to 54-year-olds accounted for 95 percent of all Gulf War-era II veterans, whereas that same age group made up about 7 in 10 nonveterans.

  4. #4 peak

    water has a higher density than oil, it will sink at the bottom of the bottle, and the oil will stay above it because it’s less dense. The same principle applies to food coloring which will also sink at the bottom of the bottle.⏎

  5. #5 peak

    uent is called __?__ .<|endoftext|>Many people believe that every word in the Bible -- including the creation account presented in Genesis -- is literally true.⏎But those same people -- especially the children among them -- often face a tough challenge when they run

  6. #6 peak

    parallels between computational and biological viruses, and the means of battling against them. Just as we humans can minimize the risks of ill health through prevention and curing, the same is true of protecting the health of our vulnerable PCs.⏎In terms of

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: this. So, then, the next thing is to think of the next thing. Now, the point is to notice that the current state. Now, the next thing to notice, the next thing to notice, the next thing to notice

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

    get the correct value from the table. For example, you select “Canada” and “VAT”, you want to get 5% in return.⏎Below this table, we want to be able to define the country in cell C11

  2. #2 peak

    of both numbers (and operators) directly below and above it. That means it begins in the same<|endoftext|>The books here address specifically how the mind learns and how this information can be applied both in academics and in life.⏎Bain, Ken.

  3. #3 peak

    use.⏎The Good News⏎Vaperma’s membrane-based Siftek, on the other hand, is expected to reduce emissions significantly. And in addition to this, it’s also supposed to be more economical. Thanks to the fact that S

  4. #4 peak

    cast ballots, one of the highest levels of participation in Venezuelan history. Incumbent President Hugo Chávez was elected for a third six-year term.Top of page⏎Venezuela had one of the highest murder rates in Latin America due,

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. L13:547 excites
  2. L7:1253 suppresses
  3. L13:1666 suppresses
  4. L11:1907 suppresses
  5. L13:2979 suppresses
  6. L13:137 suppresses
  7. L8:1322 suppresses
  8. L7:620 excites
  9. L13:1966 excites
  10. L13:1040 suppresses
  11. L7:1794 suppresses
  12. L8:915 excites
  13. L12:1583 suppresses
  14. L10:2439 excites
  15. L12:2701 excites
  16. L13:2191 excites
  17. L13:1333 suppresses
  18. L7:975 excites
  19. L12:59 excites
  20. L10:1046 suppresses
  21. L13:3053 suppresses
  22. L12:732 suppresses
  23. L13:1818 suppresses
  24. L12:2864 suppresses

Feeds later layers

  1. L15:1209 suppresses
  2. L15:1570 suppresses
  3. L15:1306 suppresses
  4. L20:2429 suppresses
  5. L17:987 suppresses
  6. L16:2247 weakly suppresses
  7. L22:2538 weakly excites
  8. L16:2523 weakly suppresses
  9. L15:2991 weakly suppresses
  10. L17:2848 weakly suppresses
  11. L27:1006 weakly suppresses
  12. L15:109 weakly suppresses
  13. L15:204 weakly suppresses
  14. L16:1572 weakly suppresses
  15. L15:654 weakly excites
  16. L17:1450 weakly suppresses
  17. L15:124 weakly suppresses
  18. L17:2011 weakly suppresses
  19. L21:337 weakly suppresses
  20. L15:2879 weakly suppresses
  21. L18:2816 weakly suppresses
  22. L15:2237 weakly excites
  23. L15:3069 weakly suppresses
  24. L25:1415 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 -2.56 and -2.441 values between -1.96 and -1.831 values between -1.83 and -1.713 values between -1.71 and -1.591 values between -1.59 and -1.472 values between -1.47 and -1.352 values between -1.35 and -1.235 values between -1.23 and -1.105 values between -1.10 and -0.987 values between -0.98 and -0.8614 values between -0.86 and -0.7414 values between -0.74 and -0.6223 values between -0.62 and -0.4939 values between -0.49 and -0.3766 values between -0.37 and -0.25200 values between -0.25 and -0.13943 values between -0.13 and -0.011678 values between -0.01 and 0.12640 values between 0.12 and 0.24173 values between 0.24 and 0.3643 values between 0.36 and 0.4812 values between 0.48 and 0.609 values between 0.60 and 0.727 values between 0.72 and 0.853 values between 0.85 and 0.974 values between 0.97 and 1.094 values between 1.09 and 1.212 values between 1.21 and 1.332 values between 1.33 and 1.451 values between 1.70 and 1.822 values between 2.19 and 2.31offnegativepositiveoften

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. uiltin
  2. inality
  3. enaries
  4. KeyId
  5. Bullet
  6. Node
  7. guilty
  8. Line
  9. -line
  10. unseren

Suppresses

  1. ensuite
  2. però
  3. その後
  4. then
  5. thereafter
  6. 的同时
  7. 副市长
  8. ますが
  9. subsequently
  10. ;br