Neuron Explorer

Qwen3-0.6B

the word 'the' after a lead-in phrase

A neuron in layer 13 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 'the' that starts a noun phrase after a short opening, such as 'what's the story' or 'in the school'. Pushing it up or down made the model's replies repetitive or off-topic.

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

    <|endoftext|>FROM ISSUE NUMBER 15 ~ SPRING 2013 GO TO TABLE OF CONTENTS⏎The Art of Limited Government⏎

  2. #2 peak

    shoulder dress.⏎So, what’s the story behind this elusive spray-on garment?⏎Who created the dress?⏎The dress technology

  3. #3 peak

    experience one or more of those problems.⏎Even five years ago, the increase in morbid obesity was noticed by pediatricians to a point that Dr.

  4. #4 peak

    said.⏎- They throw out food in damaged or outdated promotional packaging (think holiday cookies) that is still edible.⏎Farms aren’t excluded from the report’s waste list.⏎The NDRC found that growers often overplant and have

  5. #5 peak

    the current one occurred in both cases.⏎“We estimate that the probability to have such a heat or higher is generally more than two times higher today

  6. #6 peak

    women experience domestic violence⏎Two girls that Kopila supported are in the school⏎Introduction KOPILA- Nepal: ‘kopila’ means flower bud

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 have not, but I have not. The problem is, I have not. Maybe that's a problem for me to share, but the thing is, the something for example, the something for example, the problem, but the 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

    str, 'p', strlen(str));⏎printf ("'p' found at position %d.\n", pch-str+1);⏎printf ("'p' not found.\n");⏎- Compare two blocks of memory (function⏎- Locate first

  2. #2 peak

    for design projects. Projects are developed from concepts learned in Wood and Metal Design, Interior Design and/or Intro to Architecture. New technologies such as a CNC Router, CNC Plasma CAM, laser

  3. #3 peak

    iasens_), etc.<|endoftext|>The American Meteorological Society⏎) promotes the development and dissemination of information and education on the atmospheric⏎Atmospheric sciences is an umbrella term for the study of the atmosphere, its processes, the effects other systems have

  4. #4 peak

    time.⏎You should give examples from dynamic typed languages instead, e.g. clojure.⏎Here's one example:<|endoftext|>The commerce of the ancient world was

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. L11:43 strongly excites
  2. L8:7 strongly suppresses
  3. L8:10 strongly suppresses
  4. L8:4 strongly suppresses
  5. L8:16 excites
  6. L8:3 suppresses
  7. L8:14 suppresses
  8. L8:13 excites
  9. L11:2 excites unclear: blank before a year or number
  10. L5:110 excites
  11. L7:2 suppresses
  12. L7:122 suppresses
  13. L5:50 suppresses
  14. L5:84 excites
  15. L6:7 excites
  16. L6:12 suppresses
  17. L9:8 suppresses
  18. L0:1301 excites
  19. L6:3 suppresses
  20. L0:1224 excites
  21. L6:13 suppresses
  22. L8:5 excites
  23. L5:158 excites
  24. L7:110 excites

Feeds later layers

  1. L15:351 excites
  2. L14:1335 suppresses
  3. L14:1254 weakly suppresses
  4. L14:1896 weakly suppresses
  5. L14:2039 weakly suppresses
  6. L14:1572 weakly suppresses
  7. L15:1141 weakly suppresses
  8. L14:933 weakly suppresses
  9. L14:423 weakly excites
  10. L18:1366 weakly suppresses
  11. L18:688 weakly excites
  12. L14:2991 weakly suppresses
  13. L14:2372 weakly suppresses
  14. L18:2956 weakly suppresses
  15. L14:97 weakly suppresses
  16. L15:1537 weakly excites
  17. L23:638 weakly suppresses
  18. L14:644 weakly suppresses
  19. L15:2425 weakly suppresses
  20. L15:1430 weakly excites
  21. L15:611 weakly suppresses
  22. L19:1444 weakly suppresses
  23. L18:847 weakly suppresses
  24. L17:3043 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 -4.49 and -4.162 values between -3.83 and -3.503 values between -3.50 and -3.174 values between -2.84 and -2.516 values between -2.51 and -2.1821 values between -2.18 and -1.8555 values between -1.85 and -1.52178 values between -1.52 and -1.19453 values between -1.19 and -0.86852 values between -0.86 and -0.53898 values between -0.53 and -0.20721 values between -0.20 and 0.13336 values between 0.13 and 0.46127 values between 0.46 and 0.7950 values between 0.79 and 1.1215 values between 1.12 and 1.452 values between 1.45 and 1.783 values between 1.78 and 2.112 values between 2.11 and 2.441 values between 2.44 and 2.772 values between 2.77 and 3.102 values between 3.43 and 3.767 values between 3.76 and 4.0917 values between 4.09 and 4.4213 values between 4.42 and 4.7527 values between 4.75 and 5.0820 values between 5.08 and 5.4118 values between 5.41 and 5.7416 values between 5.74 and 6.0711 values between 6.07 and 6.4010 values between 6.40 and 6.736 values between 6.73 and 7.066 values between 7.06 and 7.397 values between 7.39 and 7.727 values between 7.72 and 8.056 values between 8.05 and 8.382 values between 8.38 and 8.71offnegativepositiveoften

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. engeance
  2. MOZ
  3. meld
  4. setResult
  5. étr
  6. 更新
  7. ining
  8. simult
  9. empre
  10. .ma

Suppresses

  1. individual
  2. atório
  3. цию
  4. ocab
  5. читыва
  6. вор
  7. 级别的
  8. utilus
  9. acente
  10. ...]⏎⏎