Neuron Explorer

Qwen3-0.6B

unclear: content words in how-to phrases

A neuron in layer 13 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron is unclear. It fires on a scattering of words inside practical how-to sentences, such as 'keep the end in mind', 'continuous harvest' or 'wear shockproof gloves', with no single theme. Turning it up or down gave repetitive or unrelated text.

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

    is easier to work toward a goal that seems more easily achieved. It is difficult to work towards a goal, week after week, if we don’t keep the end in mind. For example, a teen working to save $50 to have

  2. #2 peak

    75°F. Successive Sowings: Every 3 weeks until 10 to 12 weeks before your average first fall frost date for continuous harvest. Mild Climates: Sow in late fall through winter for cool–season

  3. #3 peak

    is?⏎You can access the Internet Information Services (IIS) Manager by selecting Control Panel > Administrative Tools. Select Add Website from the Connections panel, expand your host tree, right-click on Sites, and select the location of

  4. #4 peak

    helmet and bent downwards to form a hook which secured the ball of clay and candle to it. The Cornish miners would take a good supply of candles underground to last the shift.<|endoftext|>Hydrology-related problems associated with North Carolina 's major

  5. #5 peak

    pump, bail bucket if the craft is not a self bailer, a whistle and bow/stern line. It is also a good idea to be prepared with rescue ropes, throw lines, a knife and carabineers.⏎Dress properly

  6. #6 peak

    will help children get into a routine and adjust to the demands of the school day.⏎Another way parents can help their children prepare for back to school is by reviewing academic skills over the summer. This can include things like

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 will not be here. I will not be here. I will not be here. I will not be here. I will not be here. I will not be here. I will not be here. I will not be here.

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

    .⏎Why are Linux services called daemons?⏎The term was coined by the programmers of the MIT Project MAC. they took the name of maxwell's demon, an imaginary being from a thought experiment that constantly works in the background, sorting molecules.

  2. #2 peak

    actually live there (which is also another name for the God of the War). This name is strikingly similar to the south-west Khulian name for the God of Battle Rage, Oiden. Perhaps some ancient link can be inferred.⏎The

  3. #3 peak

    halosporins, the choice depending upon the patient’s health condition, medical history and severity of the disease.⏎However, Klebsiella possesses a chromosomal class A beta-lactamase giving it resistance to ampicillin. Many

  4. #4 peak

    it to screen, which is of course not the goal. So you need to redirect the output.⏎diff -u -p source.cpp.orig source.cpp > ~/patch.diff⏎~/patch.diff is here an example and you can create the file where

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. L12:55 excites
  2. L11:1044 suppresses
  3. L10:1025 excites
  4. L12:1718 excites
  5. L12:1634 suppresses
  6. L10:1297 excites
  7. L11:1525 excites
  8. L12:1577 excites
  9. L12:1869 suppresses
  10. L10:1469 suppresses
  11. L11:1988 suppresses
  12. L12:1801 excites
  13. L12:1257 suppresses
  14. L12:694 suppresses
  15. L10:2210 excites
  16. L10:867 weakly suppresses
  17. L12:2655 weakly excites
  18. L12:587 weakly excites
  19. L11:747 weakly suppresses
  20. L12:1986 weakly excites
  21. L12:2954 weakly excites
  22. L11:96 weakly excites
  23. L12:1399 weakly suppresses
  24. L9:1395 weakly suppresses

Feeds later layers

  1. L18:1427 excites
  2. L18:2623 excites
  3. L17:1528 suppresses
  4. L14:2058 suppresses
  5. L16:2965 excites
  6. L15:2096 excites
  7. L16:3069 excites
  8. L16:1970 excites
  9. L15:2562 excites
  10. L16:2472 suppresses
  11. L16:810 excites
  12. L14:2796 excites
  13. L14:257 excites
  14. L22:2784 excites
  15. L14:2637 weakly excites
  16. L16:2431 weakly suppresses
  17. L17:1938 weakly excites
  18. L15:2918 weakly suppresses
  19. L14:1413 weakly suppresses
  20. L14:576 weakly excites
  21. L18:2019 weakly suppresses
  22. L14:2168 weakly suppresses
  23. L14:2526 weakly excites
  24. L22:192 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 -2.34 and -2.261 values between -1.66 and -1.581 values between -1.41 and -1.323 values between -1.32 and -1.241 values between -1.24 and -1.157 values between -1.15 and -1.071 values between -1.07 and -0.986 values between -0.98 and -0.904 values between -0.90 and -0.824 values between -0.82 and -0.7315 values between -0.73 and -0.6518 values between -0.65 and -0.5631 values between -0.56 and -0.4836 values between -0.48 and -0.3973 values between -0.39 and -0.31123 values between -0.31 and -0.22267 values between -0.22 and -0.14582 values between -0.14 and -0.051783 values between -0.05 and 0.03568 values between 0.03 and 0.12191 values between 0.12 and 0.2075 values between 0.20 and 0.2943 values between 0.29 and 0.3727 values between 0.37 and 0.4615 values between 0.46 and 0.546 values between 0.54 and 0.634 values between 0.63 and 0.718 values between 0.71 and 0.807 values between 0.80 and 0.884 values between 0.88 and 0.963 values between 0.96 and 1.05offnegativepositiveoften

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. 未来的
  2. pcl
  3. psz
  4. .Spec
  5. LOPT
  6. Future
  7. →req
  8. Needs
  9. pc
  10. chemes

Suppresses

  1. LAB
  2. ved
  3. iphery
  4. ela
  5. _normalize
  6. ika
  7. 激活
  8. 階段
  9. 乖
  10. 窈