Neuron Explorer

Qwen3-0.6B

unclear: list commas and mixed tokens

A neuron in layer 18 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 a mix of things, including commas in lists and a few random words, and no single theme stands out. It is hard to describe in plain terms. Turning it up made the writing repeat words like "the system" or "the statement".

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

    the index. The index starts from 0 and ends at length of the array - 1.⏎i.e. the index of the first element is 0, index of the second element is 1… etc.⏎To print the

  2. #2 peak

    plain that runs from Mexico and southern Texas through Florida and along the Atlantic as far north Maryland and Delaware. It is a year-round bird within that territory but migratory farther north and farther inland. Its occurrence is also spotty farther inland because of

  3. #3 peak

    work is the "Historia General de los Hechos de los Castellanos en las Islas y Tierra Firme del Mar Océano" (General History of the deeds of the Castilians on the Islands and Mainland of the

  4. #4 peak

    to 34-year-olds had a college degree, compared with 35 percent of 35- to 44-year-olds and 38 percent of 45- to 54-year-olds.⏎Veterans from Gulf

  5. #5 peak

    Tendon Tears. These tendons are:⏎Quadriceps Tendon: The quadriceps muscle is a combination of four muscles which are the vastus lateralis, vastus medialis, vastus intermedius, and the rectus femoris

  6. #6 peak

    pronouns in Spanish agree in number and gender with the nouns they replace. In the following example, hijos is a masculine, plural noun so the masculine plural direct object pronoun los is used to replace it in the second sentence.Son muy buen

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 one that the one that the one that the one that the one that the one that the one that the one that the one that the one that the one that the one that the one that the one that the one that the one that

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

    Dan Ferguson, Gregg Garfin, Mónica Ramirez-Andreotta and Connie Woodhouse designed the Transdisciplinary Environmental Science for Society (TESS) Program. The following parts have been prepared:⏎Part 1: Fundamentals of Transdisciplinary Research (

  2. #2 peak

    in many aspects of our lives including food quality and safety. However, mounting evidence shows that accumulation of plastic waste in oceans has detrimental effects on marine wildlife. The following infographics help visualize trends in waste generation, including plastics, for 19

  3. #3 peak

    influence on the attitudes of our time.⏎In a recent interview, DiFonzo talked about his research and his fascination with this particular form of communication. The following is an edited version of that interview:⏎City: What got you interested in rumors

  4. #4 peak

    in AutoCAD and allow engineers, clients, and the public to better understand a certain design. AutoCAD is an essential visual communication tool for civil engineers.⏎The following instructions

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. L17:1160 suppresses unclear: inside numbers and names
  2. L16:2169 suppresses
  3. L17:2011 suppresses
  4. L16:1917 suppresses
  5. L17:1221 excites
  6. L14:1674 suppresses
  7. L15:1131 suppresses
  8. L14:1268 excites
  9. L16:2638 excites
  10. L8:78 excites
  11. L13:2560 suppresses
  12. L10:1062 excites
  13. L15:860 excites
  14. L4:270 excites
  15. L14:477 excites
  16. L15:2274 suppresses
  17. L15:470 excites the word 'this' after a preposition
  18. L17:1453 excites
  19. L3:16 excites
  20. L14:1254 suppresses
  21. L11:392 suppresses
  22. L14:1078 weakly suppresses
  23. L14:99 weakly excites
  24. L14:1141 weakly excites

Feeds later layers

  1. L19:1462 strongly excites
  2. L19:1737 excites
  3. L19:2384 excites
  4. L19:222 suppresses
  5. L19:2199 excites
  6. L19:2072 excites
  7. L19:1137 excites
  8. L19:1022 excites
  9. L19:687 excites
  10. L20:1988 weakly excites "the" before a topic noun
  11. L19:2586 weakly excites
  12. L20:1983 weakly excites
  13. L19:653 weakly excites
  14. L20:867 weakly suppresses
  15. L21:1119 weakly excites
  16. L19:1444 weakly suppresses
  17. L19:1557 weakly excites
  18. L19:1543 weakly excites
  19. L20:1229 weakly excites
  20. L24:2303 weakly suppresses capital G starting names
  21. L19:988 weakly excites
  22. L19:2673 weakly excites
  23. L20:2741 weakly suppresses
  24. L20:1458 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 -6.17 and -5.901 values between -5.90 and -5.632 values between -5.63 and -5.367 values between -5.36 and -5.085 values between -5.08 and -4.816 values between -4.81 and -4.545 values between -4.54 and -4.2714 values between -4.27 and -4.0023 values between -4.00 and -3.7335 values between -3.73 and -3.4632 values between -3.46 and -3.1945 values between -3.19 and -2.9270 values between -2.92 and -2.6593 values between -2.65 and -2.38115 values between -2.38 and -2.10174 values between -2.10 and -1.83167 values between -1.83 and -1.56215 values between -1.56 and -1.29288 values between -1.29 and -1.02314 values between -1.02 and -0.75382 values between -0.75 and -0.48395 values between -0.48 and -0.21446 values between -0.21 and 0.06302 values between 0.06 and 0.33204 values between 0.33 and 0.60165 values between 0.60 and 0.88129 values between 0.88 and 1.1586 values between 1.15 and 1.4259 values between 1.42 and 1.6939 values between 1.69 and 1.9631 values between 1.96 and 2.2314 values between 2.23 and 2.5017 values between 2.50 and 2.7715 values between 2.77 and 3.043 values between 3.04 and 3.313 values between 3.31 and 3.583 values between 3.58 and 3.862 values between 4.13 and 4.401 values between 4.40 and 4.67offnegativepositiveoften

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. скор
  2. iasco
  3. apsulation
  4. ")] ⏎
  5. かり
  6. ) ⏎
  7. quisition
  8. 骈
  9. arcy
  10. sto

Suppresses

  1. 罄
  2. 度
  3. ак
  4. ListComponent
  5. ness
  6. (hdc
  7. elan
  8. ivant
  9. aced
  10. lined