Neuron Explorer

Qwen3-0.6B

"to" after "found" or "said"

A neuron in layer 3 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 word "to" right after words like "found" or "said", as in "was found to be" or "was said to have". Turning it up made the writing loop and repeat, so the steering does not clearly show this pattern.

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

    both letters capitalized to distinguish it from cobalt ("Co").⏎The madwoman of Warehouses 12 & 13 named Helena G. Wells believed herself to be THE H.G. Wells the author. Since H.G. Wells - as

  2. #2 peak

    able to store subcutaneous fat. This was found to be especially the case among males. Men are more sensitive to changes in fat distributions

  3. #3 peak

    drinking water supplies of 24 major metropolitan areas—from Southern California to Northern New Jersey and from Detroit to Louisville, Ky. Most of the drug sources were found to be in the

  4. #4 peak

    the Battle of Badr. How many men did the Meccans have fighting?⏎During the heat of the Battle of Badr, the Prophet Muhammad was said to have done what?⏎Thrown sand into the air in the direction of the enemy⏎

  5. #5 peak

    established the first church in Antioch and is regarded as the founding pope of the Catholic church. Instrumental in the spread of early Christianity, Peter was said to have walked on water, witnessed the “Transfiguration of Jesus” and denied

  6. #6 peak

    much ‘party’ in such and such a village”. The coming of elections gives fresh opportunities for the crystallization of parties around patrons. Each patron may be said to have a “vote bank” which he can place at the disposal of a provincial

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 first thing. It's been a while, and the weather is getting warm. So it's been a while. So it's been a while. The first thing is just. The first thing is just. So it's been 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

    for osteopenia are lifestyle habits everyone should adopt. If you have strong bones, a healthy lifestyle can help keep them that way. If you already have osteopenia, these same habits can help reduce the risk of osteoporosis.⏎Exercise

  2. #2 peak

    processes which ultimately result in excess cost, morbidity, readmissions, and mortality.1-4⏎There are two major types of apnea:⏎- Central sleep apnea (CSA) is due to temporary failure in

  3. #3 peak

    low testosterone, but they are often written off to “stress.” Even clinical depression may be simply due to low testosterone. Men with low testosterone may also have osteopenia or even osteoporosis (thin bones). Testosterone keeps the body skin

  4. #4 peak

    thinning bones become so fragile that they are prone to fracture easily. A person with osteopenia may benefit from treatment aimed at strengthening bone.⏎How Osteopenia Happens⏎At age 30, bones are as strong and dense

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. L1:2016 weakly suppresses
  2. L2:2681 weakly suppresses
  3. L1:2336 weakly excites
  4. L0:1301 weakly suppresses
  5. L0:178 weakly suppresses
  6. L0:1224 weakly suppresses
  7. L1:2899 weakly suppresses
  8. L0:1861 weakly suppresses
  9. L2:1805 weakly suppresses
  10. L1:2241 weakly suppresses
  11. L0:2810 weakly excites
  12. L1:2011 weakly suppresses
  13. L1:1333 weakly excites
  14. L0:852 weakly excites
  15. L0:1043 weakly excites
  16. L2:483 weakly suppresses
  17. L1:2228 weakly suppresses
  18. L2:70 weakly excites
  19. L2:2868 weakly suppresses
  20. L0:2915 weakly excites
  21. L0:2703 weakly excites
  22. L0:2539 weakly excites
  23. L2:199 weakly excites
  24. L0:1896 weakly excites

Feeds later layers

  1. L8:997 excites
  2. L8:2194 excites
  3. L18:3010 weakly excites 'to' before 'be' after considered
  4. L15:321 weakly suppresses
  5. L13:2211 weakly suppresses
  6. L12:1885 weakly excites
  7. L5:847 weakly suppresses
  8. L5:724 weakly excites
  9. L14:1724 weakly suppresses
  10. L10:684 weakly suppresses
  11. L4:730 weakly excites
  12. L4:2582 weakly suppresses
  13. L10:109 weakly suppresses
  14. L8:1096 weakly excites
  15. L17:582 weakly suppresses
  16. L19:2067 weakly suppresses
  17. L6:1974 weakly excites
  18. L11:3000 weakly excites
  19. L15:400 weakly excites
  20. L4:1380 weakly suppresses
  21. L18:1975 weakly excites
  22. L10:2732 weakly excites
  23. L15:1461 weakly suppresses
  24. L8:2355 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.

2 values between -1.62 and -1.352 values between -1.35 and -1.095 values between -1.09 and -0.8214 values between -0.82 and -0.55106 values between -0.55 and -0.281681 values between -0.28 and -0.012022 values between -0.01 and 0.2656 values between 0.26 and 0.539 values between 0.53 and 0.791 values between 0.79 and 1.061 values between 1.33 and 1.601 values between 1.60 and 1.871 values between 2.14 and 2.401 values between 2.40 and 2.672 values between 2.67 and 2.941 values between 4.55 and 4.821 values between 5.09 and 5.351 values between 6.70 and 6.961 values between 8.84 and 9.11offnegativepositiveoften

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. dimension
  2. ptides
  3. ojis
  4. Brewing
  5. /method
  6. омер
  7. iedades
  8. mutations
  9. [method
  10. ��

Suppresses

  1. .Parse
  2. >manual
  3. عون
  4. esis
  5. ESIS
  6. 和其他
  7. 幹
  8. ML
  9. Nib
  10. <?=$