Neuron Explorer

Qwen3-0.6B

commas inside lists

A neuron in layer 14 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 commas that separate items in a list, usually right before the last item or before 'and'. Turning it up made the model produce loops of numbers and unrelated text, so the steering does not show a clear link to lists.

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

    subject area. These pocket-sized books are the perfect way to get ahead in a new subject quickly. Our expert authors combine facts, analysis, perspective, new ideas, and enthusiasm to make interesting and challenging topics highly readable.<|endoftext|>The MPI Multid

  2. #2 peak

    ades are one of the few animals to have survived all five of Earth’s major extinction events, and are known to tolerate extreme heat, cold, radiation, pressure, and the absence of food, water and oxygen.⏎In order to do this,

  3. #3 peak

    . You can’t find the right words, the right phrase, the right style, the right topic, the right approach, the right content, the right angle, the right thing to write. The cursor does what it does etymologically best

  4. #4 peak

    ricula. In addition, a bioenergy literacy search tool is available to help education professionals find and link bioenergy teaching resources like videos, datasets, images, articles, software and curricula to STEM based standards. The Imagine Tomorrow competition challenges teams of

  5. #5 peak

    eman, Duvendack, and Loubere, 2019), claims the article contains a worrying number of omissions, errors, inconsistencies, and that it also employs flawed methodologies. It lays out concerns that it legitimises

  6. #6 peak

    reminder that this product is for A SINGLE USER ONLY. That means that copying or sharing any part of this product with another teacher, grade level, school, department, district,

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 can't understand math, but I can't find something" " " " " " " " " " " " " " " " " " " " " " " " " " " " " " " " " " "

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

    of equipment in indoor spaces that have been sealed tightly to block out cold temperatures and wind. Symptoms of carbon monoxide exposure can include everything from headaches, dizziness and drowsiness to nausea, vomiting or tightness across the chest. Severe

  2. #2 peak

    built environment around them. Ask students to think of ways that humans influence their environment (streets, houses,<|endoftext|>Food packaging is significant for a few reasons. As a matter of first importance

  3. #3 peak

    Important Part Of The Solution⏎The more we recycle, the less garbage winds up in our landfills and incineration plants. By reusing aluminum, paper, glass, plastics, and other materials, we can save production and energy costs,

  4. #4 peak

    ⏎What does STEM have to do with Wall Street?⏎When you think about CAREERS on Wall Street, what usually comes to mind are Brokers, Traders, Investment Bankers, and other jobs typically filled by those with degrees in Finance and

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. L13:2931 excites
  2. L13:264 excites
  3. L13:1789 excites
  4. L13:592 suppresses
  5. L13:2742 suppresses
  6. L11:116 excites
  7. L10:163 suppresses
  8. L13:749 weakly excites
  9. L12:1243 weakly excites
  10. L13:926 weakly excites
  11. L13:1938 weakly suppresses
  12. L8:1104 weakly suppresses
  13. L9:2726 weakly excites
  14. L12:2057 weakly suppresses
  15. L12:1235 weakly suppresses commas inside numbers and lists
  16. L13:199 weakly excites
  17. L0:1633 weakly excites
  18. L12:2981 weakly suppresses
  19. L13:767 weakly suppresses
  20. L10:499 weakly excites
  21. L13:2641 weakly suppresses
  22. L13:1020 weakly suppresses
  23. L0:2199 weakly excites
  24. L13:2625 weakly excites

Feeds later layers

  1. L15:2314 suppresses
  2. L16:798 suppresses
  3. L19:1414 suppresses
  4. L16:1139 excites 'and' in lists of places or names
  5. L16:913 excites
  6. L16:2327 weakly suppresses
  7. L16:2279 weakly suppresses
  8. L15:677 weakly suppresses
  9. L16:1809 weakly suppresses comma inside a list of names or acronyms
  10. L16:852 weakly excites
  11. L15:2979 weakly suppresses
  12. L20:1114 weakly excites
  13. L15:3020 weakly excites
  14. L15:2525 weakly suppresses
  15. L16:1163 weakly excites
  16. L16:2360 weakly suppresses
  17. L18:1207 weakly suppresses
  18. L20:610 weakly suppresses
  19. L16:2922 weakly excites
  20. L16:1749 weakly suppresses
  21. L16:634 weakly suppresses
  22. L15:1504 weakly excites
  23. L21:2676 weakly suppresses
  24. L18:1693 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 -2.60 and -2.491 values between -1.83 and -1.723 values between -1.61 and -1.501 values between -1.50 and -1.394 values between -1.39 and -1.283 values between -1.17 and -1.065 values between -1.06 and -0.955 values between -0.95 and -0.845 values between -0.84 and -0.7315 values between -0.73 and -0.6211 values between -0.62 and -0.5119 values between -0.51 and -0.4037 values between -0.40 and -0.29148 values between -0.29 and -0.18585 values between -0.18 and -0.071620 values between -0.07 and 0.04962 values between 0.04 and 0.15337 values between 0.15 and 0.2696 values between 0.26 and 0.3726 values between 0.37 and 0.486 values between 0.48 and 0.595 values between 0.59 and 0.702 values between 0.70 and 0.812 values between 0.81 and 0.921 values between 0.92 and 1.032 values between 1.03 and 1.142 values between 1.14 and 1.251 values between 1.36 and 1.481 values between 1.48 and 1.591 values between 1.70 and 1.81offnegativepositiveoften

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. prez
  2. (pre
  3. pres
  4. PRE
  5. сов
  6. PRES
  7. ewing
  8. (ip
  9. 瓜
  10. 三种

Suppresses

  1. unky
  2. &)⏎
  3. 读后
  4. oky
  5. çı
  6. hti
  7. xlsx
  8. 小时前
  9. licken
  10. achelor