Neuron Explorer

Qwen3-0.6B

blank spot where a clock time comes next

A neuron in layer 23 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 fires on the empty space right after a date or a 'at' phrase, just where a clock time such as 4:39 PM or 2:00 p.m. is expected next. Turning it up did not produce a clear theme in what the model wrote.

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

    Previous Article:||World match.|⏎|Next Article:||Ukraine's Orange Revolution.|<|endoftext|>posted on November 17, 2014 23:23⏎Scientists from the WCS (Wildlife Conservation Society),

  2. #2 peak

    .com⏎Fri Sep 10 00:10:22 CEST 2010⏎On 9/9/10 4:39 PM, Baba wrote:⏎> In below code "

  3. #3 peak

    AdminStaff1 · Views: 1,130⏎Contributed by: AdminStaff1 · April 10, 2008 @ 12

  4. #4 peak

    pediatric and adult hematologists. This common disease will be explored in depth in an Education Program Session today at 7:30 a.m. and again at 2:00 p.m. in rooms 393-39

  5. #5 peak

    photos and video footage from NASA satellites to discuss the visual phenomenon known as the Northern Lights.⏎- Sat., Mar. 2, 2013, 10:30 AM⏎- Terrace Gallery⏎Please use the event calendar

  6. #6 peak

    .⏎Though, I guess, it's not the first time it's happened...<|endoftext|>Last updated on September 17th, 2022 at 02:16 pm⏎Synodontis catfish are one of the

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: the more local, the more local, the more local, the more local... new local and new local and new local and new local and new local... and I am thinking of a new local to add. I need to create 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

    148 ft (350 m)|⏎|Time zone||Eastern (EST) (UTC-5)|⏎|• Summer (DST)||EDT (UTC-4)|⏎|GNIS feature ID||097

  2. #2 peak

    'tray' and covering it with paper.⏎My cousin was 24 years old on Friday April 5th in 1974. On what day⏎of the week was she born?⏎A tetromino is made up

  3. #3 peak

    11)|| 0.805⏎very high · 43rd⏎|Currency||Euro (€) (EUR)|⏎|Time zone||EET (UTC+2)|⏎|-||Summer (DST)||E

  4. #4 peak

    Windows 11 users will benefit hugely from learning how to zip their bulky files and if the need arises to unzip them.⏎Also read: How To Change Date And Time In Windows 11⏎How To Zip Your Files In Windows 1

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. L20:1335 suppresses
  2. L22:2068 suppresses
  3. L19:2242 suppresses
  4. L9:811 suppresses
  5. L21:1185 excites
  6. L20:1355 suppresses
  7. L20:2792 excites
  8. L12:2338 excites
  9. L21:375 suppresses
  10. L22:2982 excites
  11. L22:2445 excites
  12. L5:55 suppresses
  13. L4:2826 excites
  14. L16:1668 suppresses
  15. L21:2290 excites
  16. L14:1088 excites
  17. L4:2570 suppresses
  18. L22:1536 weakly suppresses
  19. L17:1872 weakly excites
  20. L4:586 weakly suppresses
  21. L9:1057 weakly excites
  22. L20:1155 weakly excites
  23. L18:1511 weakly suppresses
  24. L4:2497 weakly suppresses

Feeds later layers

  1. L27:1288 suppresses
  2. L24:637 suppresses
  3. L25:1832 suppresses the word 'local'
  4. L24:1904 weakly suppresses
  5. L24:950 weakly suppresses
  6. L24:2654 weakly excites
  7. L27:679 weakly suppresses
  8. L26:2174 weakly suppresses
  9. L27:1207 weakly suppresses
  10. L27:710 weakly suppresses
  11. L25:1021 weakly suppresses
  12. L24:1023 weakly excites
  13. L24:1069 weakly excites
  14. L27:391 weakly excites
  15. L26:2435 weakly suppresses
  16. L25:2922 weakly suppresses
  17. L24:3022 weakly excites
  18. L26:2610 weakly excites
  19. L24:188 weakly excites
  20. L25:1872 weakly excites
  21. L24:1463 weakly excites
  22. L25:224 weakly excites
  23. L25:155 weakly suppresses
  24. L26:1274 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 -10.85 and -9.701 values between -8.54 and -7.392 values between -5.07 and -3.924 values between -3.92 and -2.7611 values between -2.76 and -1.61225 values between -1.61 and -0.453555 values between -0.45 and 0.7161 values between 0.71 and 1.869 values between 1.86 and 3.0213 values between 3.02 and 4.177 values between 4.17 and 5.334 values between 5.33 and 6.485 values between 6.48 and 7.641 values between 7.64 and 8.803 values between 9.95 and 11.112 values between 13.42 and 14.571 values between 16.88 and 18.041 values between 26.13 and 27.291 values between 34.22 and 35.38offnegativepositiveoften

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. liğini
  2. iameter
  3. ApplicationBuilder
  4. /embed
  5. progressbar
  6. راء
  7. xAB
  8. xAA
  9. 台上
  10. iores

Suppresses

  1. Local
  2. local
  3. local
  4. Local
  5. 本地
  6. _local
  7. →Local
  8. →local
  9. -local
  10. LOCAL