Neuron Explorer

Qwen3-0.6B

'least' in 'at least' hedges

A neuron in layer 11 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

Fires on the word 'least' when it follows 'at' in phrases like 'at least not yet' or 'or at least the elder males'. These are hedges that qualify a claim. Turning it up did not produce hedging in the writing, so the effect on output is unclear.

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

    new research project at MIT called Sparsh is aiming to fix that oversight.⏎Sparsh (the Hindi word for "touch") isn't an app, at least not in the way we generally use the word. It's a tool that's

  2. #2 peak

    ’s no way around the laws of thermodynamics, at least not yet. Quality of food is an important issue as well, don’t get

  3. #3 peak

    homeschool. Along the way, we have purposely sought out resources and opportunities to develop fluency in a second language.⏎Finding materials for Norwegian is not easy (at least where I live) so I was very excited to discover the New Amigos board

  4. #4 peak

    the Industrial Revolution (XVIII century) and the golden age of British education throughout the Victorian era (XIX century).⏎With the information that training, and particularly science schooling, is an endeavor that relies not solely on what occurs in the classroom

  5. #5 peak

    regardless of whether or not they will be able to fill that extra time with something profound. In a state of nature, people are less concerned with death (at least as an abstract entity, obviously the predator lurking in the bushes is a more real

  6. #6 peak

    tool that writes your application - your code is a set of instructions to the compiler about what you want it to do. Since compilers aren't very creative (at least we hope) you have to be very precise, and the compiler tends to be

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 difference between ... ... ... This is a great article. ... ... ... ... It supports the view that ... ... ... Based on the text message ... ... This is peer-to-peer (PMMSO) ... ... ... ... 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

    (2009). The New Physics for the Twenty-First Century. Cambridge University Press.⏎C. W. Kilmister (1994/2005). Eddington's Search for a Fundamental Theory: A

  2. #2 peak

    clearly indicates that it belongs to the earliest period at Makkah. Traditions also show that the revelations were suspended for a time, which caused the Holy Prophet (upon whom be Allah's peace) to be deeply distressed and grieved.

  3. #3 peak

    : Krista Tippett⏎Managing Producer: Kate Moos⏎Associate Producer: Nancy Rosenbaum⏎Associate Producer: Shubha Bala⏎Technical Director/Producer: Chris Heagle⏎Senior Editor: Trent Gilliss⏎Associate Web Developer

  4. #4 peak

    species as having “orthographic” abilities.⏎In the experiment, pigeons were trained to peck four-letter English words as they came up on a screen, or to instead peck a symbol when a four-letter non-word, such as “

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. L9:2010 suppresses
  2. L10:1459 suppresses
  3. L10:135 excites 'or' in alternative requirements
  4. L6:1373 weakly excites
  5. L0:1286 weakly excites
  6. L10:144 weakly suppresses
  7. L7:2716 weakly excites
  8. L10:89 weakly suppresses
  9. L7:625 weakly excites
  10. L2:493 weakly suppresses
  11. L10:322 weakly excites
  12. L2:2440 weakly suppresses
  13. L10:2650 weakly suppresses
  14. L1:2568 weakly excites
  15. L2:319 weakly suppresses 'of' after 'pattern' or 'patterns'
  16. L7:428 weakly excites
  17. L10:2535 weakly excites
  18. L9:464 weakly suppresses
  19. L4:108 weakly excites "or" offering a choice in web text
  20. L10:1124 weakly suppresses
  21. L5:200 weakly excites
  22. L9:1624 weakly suppresses
  23. L1:2890 weakly suppresses
  24. L8:765 weakly excites

Feeds later layers

  1. L16:1124 suppresses
  2. L18:2557 weakly excites
  3. L14:2276 weakly excites opening bracket after a term or number
  4. L14:2820 weakly excites "and" or "or" linking two counted items
  5. L25:2471 weakly suppresses
  6. L15:1501 weakly excites "and" joining two parts of a name or phrase
  7. L15:2010 weakly suppresses
  8. L12:969 weakly suppresses
  9. L12:1452 weakly excites
  10. L15:2993 weakly suppresses
  11. L14:2784 weakly excites
  12. L14:1256 weakly suppresses
  13. L16:215 weakly excites
  14. L20:1750 weakly suppresses
  15. L14:2595 weakly excites
  16. L17:727 weakly excites
  17. L21:750 weakly suppresses
  18. L17:1318 weakly excites
  19. L15:1189 weakly suppresses
  20. L18:1133 weakly excites
  21. L16:2049 weakly excites
  22. L14:2972 weakly suppresses
  23. L13:2923 weakly excites
  24. L16:1454 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 -2.34 and -2.071 values between -1.53 and -1.262 values between -1.26 and -0.995 values between -0.99 and -0.7215 values between -0.72 and -0.45369 values between -0.45 and -0.182763 values between -0.18 and 0.09688 values between 0.09 and 0.3638 values between 0.36 and 0.638 values between 0.63 and 0.905 values between 0.90 and 1.173 values between 1.17 and 1.442 values between 1.44 and 1.712 values between 2.25 and 2.521 values between 3.87 and 4.141 values between 4.68 and 4.951 values between 4.95 and 5.221 values between 7.65 and 7.921 values between 8.19 and 8.46offnegativepositiveoften

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. apest
  2. เฉพาะ
  3. ippers
  4. 鲘
  5. antic
  6. ArrayOf
  7. AsyncResult
  8. any
  9. apol
  10. unch

Suppresses

  1. oid
  2. ishes
  3. N
  4. aids
  5. 憾
  6. オ
  7. oi
  8. ftime
  9. 各
  10. alphabet