Neuron Explorer

Qwen3-0.6B

numbers and quantities mid-phrase

A neuron in layer 11 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 fires on the spot right before a number or amount, like 'a few thousand', '6,546 farmers' or '93 killed'. It seems to expect a quantity to come next. Steering did not show a clear effect on the writing.

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

    regular structure may be inherited after Recombination. For example, primordial magnetic fields may be responsible for very large scale filaments ( 100 Mpc) as discussed later. Moreover, the existence of periodic structures forming a lattice,

  2. #2 peak

    high concentration of earthquakes. What country(ies) are nearby? What type of boundary are the quakes on?⏎Your paper should meet the following requirements:⏎3-4 pages in length (does not include title and reference page)⏎1-2

  3. #3 peak

    couldn't believe them. I had a great guilt feeling when I actually found out about what happened in these camps. I had talked in terms of possibly a few thousand having been murdered, but thinking in terms of six million

  4. #4 peak

    miles to the north.⏎Each boat is a monument to human tenacity — or desperation. So is Joal. Fifty years ago, it was a village of 6,546 millet farmers and five pirogues. Then

  5. #5 peak

    well and is hinted at in this illustration. The total amount of material entrained in the broad flow is at least 400 times the mass of the Sun per year. Note that a more localized, narrower jet is shown, this jet

  6. #6 peak

    ⏎The apple is the most diverse food plant in the world. According to food journalist and apple expert Rowan Jacobsen, there used to be 16,000 types of apples in the U.S. alone. A current review

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 this, and I have been thinking about this, and I have been thinking about this. The latest achievement, the latest achievement, the latest achievement. The latest achievement. The latest achievement. The latest achievement. I

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

    .⏎Visit our Ed Clark Museum of Missouri Geology on Geologic Map Day to learn more about geologic maps. Open weekdays from 8 a.m. to 4:30 p.m., we are located at 111 Fair

  2. #2 peak

    24 hour clock, at certain times, all the digits are⏎consecutive. How many times like this are there between midnight⏎and 7 a.m.?⏎These practical challenges are all about making a 'tray' and covering

  3. #3 peak

    quantity of consumption never changes on price fluctuations, that push consumption oil producers to extreme profit zone. Decrease in prices would put them under normal profits. Any additional cost is passed on to customers, eventually companies being the fortune getters.⏎One other thing

  4. #4 peak

    are its causes? This article from HealthHearty discusses how it impinges on our health and how it can be treated.⏎Blood pressure is a measure of the force exert

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. L9:43 excites
  2. L10:212 excites
  3. L9:566 excites
  4. L6:442 excites
  5. L10:435 suppresses
  6. L10:1539 excites
  7. L10:2280 excites
  8. L6:1849 excites
  9. L9:365 excites
  10. L7:1524 suppresses
  11. L7:171 excites
  12. L7:830 suppresses
  13. L7:1742 suppresses
  14. L9:183 suppresses
  15. L10:2300 suppresses
  16. L8:1925 suppresses
  17. L10:2012 excites
  18. L8:1939 suppresses
  19. L8:853 excites slot right before a quantity
  20. L10:127 weakly suppresses
  21. L4:311 weakly excites
  22. L9:975 weakly excites
  23. L8:2242 weakly suppresses
  24. L10:2527 weakly suppresses

Feeds later layers

  1. L13:1635 suppresses
  2. L15:2421 suppresses
  3. L16:556 excites
  4. L12:2195 weakly excites
  5. L15:1423 weakly excites
  6. L17:453 weakly excites
  7. L18:2519 weakly excites
  8. L14:840 weakly excites
  9. L13:2209 weakly excites
  10. L13:852 weakly excites
  11. L21:1439 weakly suppresses
  12. L18:585 weakly excites
  13. L12:729 weakly excites
  14. L19:758 weakly excites
  15. L18:1429 weakly excites
  16. L21:382 weakly excites
  17. L12:263 weakly excites
  18. L13:1684 weakly suppresses
  19. L16:802 weakly excites
  20. L18:2718 weakly suppresses
  21. L15:2209 weakly suppresses person names and role nouns
  22. L13:1487 weakly suppresses
  23. L13:2558 weakly excites unclear: quantities and measured changes
  24. L24:1486 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 -2.16 and -1.991 values between -1.82 and -1.651 values between -1.65 and -1.481 values between -1.48 and -1.315 values between -1.31 and -1.149 values between -0.97 and -0.8014 values between -0.80 and -0.6328 values between -0.63 and -0.4674 values between -0.46 and -0.29287 values between -0.29 and -0.122333 values between -0.12 and 0.05747 values between 0.05 and 0.22160 values between 0.22 and 0.3972 values between 0.39 and 0.5641 values between 0.56 and 0.7327 values between 0.73 and 0.9019 values between 0.90 and 1.0720 values between 1.07 and 1.2417 values between 1.24 and 1.4111 values between 1.41 and 1.588 values between 1.58 and 1.757 values between 1.75 and 1.924 values between 1.92 and 2.095 values between 2.09 and 2.267 values between 2.26 and 2.432 values between 2.43 and 2.602 values between 2.77 and 2.942 values between 2.94 and 3.111 values between 3.11 and 3.281 values between 3.62 and 3.791 values between 4.47 and 4.64offnegativepositiveoften

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. 折扣
  2. деся
  3. 减免
  4. nights
  5. Vs
  6. FFECT
  7. Machine
  8. max
  9. Garn
  10. Ve

Suppresses

  1. ptive
  2. แพง
  3. ObjectId
  4. )?$
  5. guts
  6. ョン
  7. Agility
  8. 墕
  9. 骞
  10. .omg