Neuron Explorer

Qwen3-0.6B

'take time to' duration phrases

A neuron in layer 12 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 on the words 'to', 'take' and 'time' in sentences about how long something needs, such as 'it will take some time to work' or 'up to a month to appear'. Turning it up made the model talk about waiting and things taking time.

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

    physical product. After purchase, you will be provided a one time link to download ebooks to your computer. Orders paid by PayPal require up to 8 business hours to verify payment and release electronic media. For immediate downloads, payment with credit card is

  2. #2 peak

    daughter to make academics a high priority beginning with his or her freshman year. It is important to remember that it may take him or her a while to adjust to the rigorous academic demands of college.⏎4. Obtaining Marketable Skills⏎

  3. #3 peak

    a homeopathic medicine stimulates your own internal healing mechanisms and while these are in action there is no need to give more. We wait and watch while the healing process takes place. There may be a slight worsening of symptoms (aggravation) at

  4. #4 peak

    surrounding their spinal cords and brains.⏎Eleven of those people have died. The case count rises daily, as symptoms of fungal infection can take up to a month to appear, and there's often a delay in case reporting.⏎Most of the

  5. #5 peak

    this phase of your treatment. The gaps can be anywhere between two weeks to a month. At this point, you may see a decrease in symptoms. It can take up to a year on your maintenance dose to notice a true change. If the

  6. #6 peak

    to awake upon feeling the urge to urinate before having a bedwetting accident. This treatment does take some time to work, many weeks in some individuals, and requires perseverance and commitment. It is

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 been thinking about the same problem, and I have had the same problem, and now, I can’t see it, and I have had a problem, and now, I can’t see it, and I have had a

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

    of the storm.⏎It will take about two days to prepare the shuttle to move off the launch pad, and on Sunday it appeared that some of those preparations had already begun. The crawler transporter that takes the shuttle to (and from)<|endoftext|>German

  2. #2 peak

    of problems that are less obvious, but more insidious. First, there are “slow-burn” problems that may not receive enough attention—until it's too late. Think: drug-resistant “superbugs.” Second, there are emerging technologies

  3. #3 peak

    touching only the ties/straps.⏎8. After removing the face covering, always wash your hands with soap and water for at least 20 seconds or use hand sanitiser with a minimum of 60% alcohol base.⏎9.

  4. #4 peak

    who care about diversity, equity and justice can find news, suggestions, conversation and support.⏎The big day is just around the corner—but it’s not too soon to think about the days after.⏎Blended learning can transform a classroom and give a

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. L7:2577 excites
  2. L8:2192 excites
  3. L9:2326 excites
  4. L11:2204 suppresses
  5. L8:2721 excites
  6. L8:2749 excites
  7. L10:1725 excites
  8. L10:2599 weakly excites
  9. L10:2683 weakly suppresses
  10. L8:1536 weakly excites
  11. L11:1737 weakly suppresses
  12. L3:1151 weakly suppresses
  13. L6:1260 weakly excites
  14. L10:1974 weakly suppresses
  15. L11:151 weakly suppresses
  16. L9:3035 weakly excites
  17. L2:1082 weakly suppresses
  18. L1:268 weakly suppresses
  19. L5:1955 weakly excites
  20. L4:2080 weakly suppresses
  21. L11:2317 weakly excites
  22. L7:1643 weakly suppresses
  23. L10:3013 weakly excites
  24. L2:1248 weakly suppresses

Feeds later layers

  1. L19:1917 excites
  2. L15:2629 excites
  3. L22:2932 excites
  4. L20:2982 excites
  5. L17:2964 excites
  6. L17:1512 excites
  7. L15:1520 excites
  8. L24:906 excites
  9. L19:1712 excites
  10. L22:1074 excites
  11. L18:2199 excites
  12. L18:1546 excites
  13. L21:2955 excites
  14. L18:1505 excites
  15. L17:2292 excites
  16. L18:1750 excites the blank slot before a year
  17. L16:583 excites
  18. L18:1125 excites
  19. L17:2750 excites
  20. L17:2446 excites
  21. L14:1778 excites
  22. L20:1155 excites
  23. L21:1284 excites
  24. L22:486 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.

1 values between -2.07 and -1.933 values between -1.37 and -1.232 values between -1.23 and -1.093 values between -1.09 and -0.953 values between -0.95 and -0.819 values between -0.81 and -0.678 values between -0.67 and -0.5324 values between -0.53 and -0.39100 values between -0.39 and -0.25480 values between -0.25 and -0.101788 values between -0.10 and 0.041039 values between 0.04 and 0.18329 values between 0.18 and 0.3283 values between 0.32 and 0.4617 values between 0.46 and 0.606 values between 0.60 and 0.742 values between 0.74 and 0.886 values between 0.88 and 1.022 values between 1.16 and 1.301 values between 1.58 and 1.721 values between 2.14 and 2.281 values between 3.41 and 3.55offnegativepositiveoften

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. wait
  2. waiting
  3. 等待
  4. delay
  5. đợi
  6. 時間が
  7. 漫长
  8. 较长
  9. chờ
  10. (wait

Suppresses

  1. aise
  2. ycastle
  3. yle
  4. 未经授权
  5. ái
  6. orde
  7. 何必
  8. рова
  9. anced
  10. przec