Neuron Explorer

Qwen3-0.6B

digits inside years like 19xx

A neuron in layer 5 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 is weakly active on the middle digits of years written out in full, such as the '8' in 1986 or the '7' in 1972. Its strongest examples are only a little above the typical range, and turning it up gave scrambled text with no clear link to years.

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

    .⏎In the Congo-Kinshasa Province, the Sisters believe that protection of the environment is an essential and

  2. #2 peak

    aint-Laurent, known as Yves Saint Laurent was a French fashion designer, one of the greatest names in French fashion

  3. #3 peak

    s would have said racism didn’t exist, but the progress was palpable and undeniable.⏎In 1986, “The Cosby Show” was the most popular show on television, but

  4. #4 peak

    - THE CREATIVE TEAM⏎- THE CREATIVE ARTISTS⏎- IMPLEMENTING THE DESIGNS⏎- PRODUCTIONG MANAGEMENT / STAGE MANAGEMENT / TECHNICAL DIRECTION⏎- PRODUCTIONG CREWS⏎

  5. #5 peak

    Jerusalem on the shroud. The weave of the linen has also been linked with first-century Syria.⏎In 1988, three laboratories in Oxford, Zurich and Tucson each used radiocarbon dating

  6. #6 peak

    commemorating the electrification scheme in 192<|endoftext|>David Bourget (Western Ontario)⏎David Chalmers (ANU, NYU)⏎Rafael De Clercq⏎Jack Alan Reynolds⏎Learn more about PhilPapers⏎O

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 a problem with the same old one. I have to think about it again. It is the same old problem with a new problem. The same old problem is something I have been thinking about. I think I have to think

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

    were collected, studied, and enjoyed widely in the eighteenth century. Interest in collecting reproductive prints stimulated new developments in printmaking techniques, including the mezzotint process and color printing. In a domestic setting, prints might be assembled in a

  2. #2 peak

    with a curved circular path was framed by rhododendrons and a see-through vault of birch. Beyond the entrance gate, the nave began with a broad central path, flanked by ‘columns’ of ascending flowering cherries.⏎

  3. #3 peak

    bee. Covering his more than 20-year career, Bright Common Spikes: The Sculpture of John Bisbee, features 20 large-scale abstract sculptures of nails that are welded and forged into various organic configurations. The exhibition will

  4. #4 peak

    crowds that hadn't eaten in three days. I can still see Schim [sic], a wisp of a man, standing on the jeep silhouetted against the darkling sky quietly photographing

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. L3:202 excites
  2. L3:366 excites
  3. L3:993 suppresses
  4. L3:99 excites
  5. L3:855 excites
  6. L3:1303 weakly suppresses
  7. L3:204 weakly suppresses
  8. L4:83 weakly excites
  9. L3:950 weakly excites
  10. L2:695 weakly excites
  11. L3:1380 weakly suppresses
  12. L3:550 weakly excites
  13. L2:1043 weakly suppresses
  14. L4:2326 weakly excites
  15. L4:1201 weakly excites
  16. L4:914 weakly excites
  17. L4:322 weakly excites
  18. L3:2988 weakly excites
  19. L1:1703 weakly excites
  20. L4:84 weakly excites
  21. L3:836 weakly suppresses
  22. L1:928 weakly suppresses
  23. L4:1793 weakly suppresses
  24. L0:1866 weakly excites

Feeds later layers

  1. L13:42 suppresses
  2. L12:33 suppresses
  3. L8:1803 excites
  4. L22:2520 excites
  5. L16:480 weakly suppresses
  6. L14:8 weakly suppresses
  7. L6:634 weakly excites
  8. L16:203 weakly suppresses
  9. L16:125 weakly suppresses unclear: mid-word and mid-phrase fragments
  10. L6:964 weakly excites
  11. L21:846 weakly excites
  12. L12:11 weakly suppresses
  13. L16:1308 weakly excites
  14. L8:436 weakly excites
  15. L16:444 weakly suppresses
  16. L6:1928 weakly excites
  17. L15:17 weakly suppresses
  18. L21:2466 weakly excites
  19. L6:1030 weakly excites
  20. L10:11 weakly suppresses
  21. L7:1029 weakly excites
  22. L24:804 weakly excites
  23. L17:2378 weakly excites 'yes' answering a question
  24. L6:18 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.

1 values between -1.96 and -1.872 values between -1.44 and -1.354 values between -1.26 and -1.183 values between -1.18 and -1.093 values between -1.09 and -1.003 values between -1.00 and -0.914 values between -0.91 and -0.835 values between -0.83 and -0.745 values between -0.74 and -0.659 values between -0.65 and -0.5717 values between -0.57 and -0.4819 values between -0.48 and -0.3943 values between -0.39 and -0.3152 values between -0.31 and -0.22177 values between -0.22 and -0.13539 values between -0.13 and -0.051884 values between -0.05 and 0.04749 values between 0.04 and 0.13215 values between 0.13 and 0.2184 values between 0.21 and 0.3042 values between 0.30 and 0.3917 values between 0.39 and 0.4711 values between 0.47 and 0.565 values between 0.56 and 0.651 values between 0.65 and 0.737 values between 0.73 and 0.822 values between 0.82 and 0.913 values between 0.91 and 0.991 values between 1.17 and 1.251 values between 1.43 and 1.51offnegativepositiveoften

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. setC
  2. _attached
  3. Opens
  4. CAF
  5. flexDirection
  6. 嵴
  7. legalization
  8. caf
  9. Platforms
  10. 렛

Suppresses

  1. issant
  2. agan
  3. iss
  4. essim
  5. 利
  6. 团
  7. 相当于
  8. 联想
  9. issan
  10. 分