Neuron Explorer

Qwen3-0.6B

unclear: digits inside amounts and figures

A neuron in layer 11 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 a digit or word in the middle of a number, price or quantity, such as the zeros in a large sum of money. It turns negative on numbers written as percentages or fractions. Turning it up made the model drift into arithmetic word problems, though the link is loose.

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

    maturity.⏎Let's take Bond A, issued by Company Alpha, as an example. It has the following data:⏎- Face value: $1,000⏎- Annual coupon rate: 5%⏎- Coupon Frequency: Annual⏎-

  2. #2 peak

    outcome of $400,000 compared to an average teacher.⏎Effect of Good Teachers on the Economy⏎What

  3. #3 peak

    decrease the body's ability to use carbs for key functions. The USDA's Dietary Reference Intakes recommend getting from 45 to 65 percent of daily calories - from 900 to 1,300 calories in a

  4. #4 peak

    , over a second pulley wheel in the fixed block, and back to the free block, the mechanical advantage is 3. A 300-lb load can be raised by a pull of 100 lb on the free

  5. #5 peak

    idea been as popular with collectors and taxonomists of these groups in the past, the current number could well be in excess of 100,000 species. Recent estimates suggest that there may be more than 500

  6. #6 peak

    , and not stopped climbing.⏎According to U.S. Centers for Disease Control and Prevention data, opioids accounted for 75% of the 100,000 overdose deaths in the last 12 months,

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: - something about a place, a person, a country, a region, of a specific name of a country or existing in a local language of a specific city of a country or existing in a region of a specific

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

    that represents a fraction of the number 100.⏎In other words, it is a part of a general number that should be considered divided by 100 units. The symbol with which this fraction is represented is ” % “.⏎

  2. #2 peak

    tolerance⏎To use the formula, first figure out what you want your error of tolerance to be. For example, you may be happy with a confidence level of 95 percent (giving a margin error of 0.05),

  3. #3 peak

    % is the same as 25/100 and equivalent to 0.25 . One of the most common uses of

  4. #4 peak

    converted to a percentage such as 0.10 percent, which indicates that one-tenth of one percent of a person’s blood is alcohol. The amount of alcohol it takes to reach

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. L1:925 weakly suppresses
  2. L6:1287 weakly excites
  3. L8:2738 weakly suppresses
  4. L7:1428 weakly suppresses
  5. L7:1945 weakly excites
  6. L6:1573 weakly excites
  7. L10:1587 weakly excites
  8. L10:2318 weakly suppresses
  9. L1:2654 weakly excites
  10. L7:2595 weakly suppresses
  11. L3:2310 weakly excites
  12. L1:1591 weakly excites
  13. L8:5 weakly suppresses
  14. L0:951 weakly excites
  15. L8:110 weakly excites
  16. L6:2316 weakly suppresses
  17. L4:2659 weakly excites
  18. L10:1485 weakly suppresses
  19. L2:1998 weakly suppresses
  20. L5:1895 weakly suppresses
  21. L5:1694 weakly suppresses
  22. L7:1888 weakly suppresses
  23. L6:1547 weakly suppresses
  24. L1:2996 weakly suppresses

Feeds later layers

  1. L12:2728 suppresses
  2. L12:2823 excites
  3. L12:2236 excites
  4. L12:322 excites
  5. L13:1130 suppresses
  6. L24:316 suppresses
  7. L15:1724 weakly suppresses
  8. L12:2195 weakly suppresses
  9. L13:1629 weakly suppresses
  10. L14:1863 weakly excites
  11. L12:1800 weakly excites
  12. L12:2116 weakly suppresses
  13. L12:625 weakly excites
  14. L22:2133 weakly suppresses
  15. L13:3000 weakly excites
  16. L12:2002 weakly excites
  17. L27:1325 weakly suppresses
  18. L15:2534 weakly excites
  19. L23:1224 weakly suppresses
  20. L23:1101 weakly suppresses
  21. L15:708 weakly suppresses
  22. L18:1977 weakly suppresses
  23. L13:2810 weakly excites
  24. L12:2664 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 -0.67 and -0.611 values between -0.61 and -0.563 values between -0.56 and -0.506 values between -0.50 and -0.4418 values between -0.44 and -0.3938 values between -0.39 and -0.3371 values between -0.33 and -0.27114 values between -0.27 and -0.22209 values between -0.22 and -0.16289 values between -0.16 and -0.10426 values between -0.10 and -0.05662 values between -0.05 and 0.01559 values between 0.01 and 0.07502 values between 0.07 and 0.13382 values between 0.13 and 0.18250 values between 0.18 and 0.24160 values between 0.24 and 0.30110 values between 0.30 and 0.3554 values between 0.35 and 0.4117 values between 0.41 and 0.4710 values between 0.47 and 0.529 values between 0.52 and 0.584 values between 0.58 and 0.642 values between 0.64 and 0.692 values between 0.92 and 0.982 values between 0.98 and 1.041 values between 1.09 and 1.151 values between 1.21 and 1.261 values between 1.26 and 1.321 values between 1.43 and 1.492 values between 1.55 and 1.60offnegativepositiveoften

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. .base
  2. base
  3. base
  4. (base
  5. -base
  6. _base
  7. atre
  8. →base
  9. etre
  10. uminum

Suppresses

  1. че
  2. ://{
  3. FormGroup
  4. acked
  5. record
  6. ⏎
  7. FromArray
  8. ngrx
  9. FlatButton
  10. illez