Neuron Explorer

Qwen3-0.6B

dictionary and citation boilerplate

A neuron in layer 9 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 the 'by' and 'for' in copyright and source notes, like 'from WordNet 3.0 Copyright 2006 by Princeton University'. It goes negative on sales and price notices. Turning it up made the model write everyday mixed text, and turning it down made it repeat words.

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

    A 8|⏎Article from a library⏎|Web site with no author|⏎The examples above are reproduced from https://owl.english.purdue.edu/owl/resource/747/22/⏎

  2. #2 peak

    tionary, Creative Commons Attribution/Share-Alike License⏎- n. An underwater explorer.⏎from WordNet 3.0 Copyright 2006 by Princeton University. All rights reserved.⏎- n. an underwater swimmer equipped with

  3. #3 peak

    avia and Finland with the Kola Peninsula in Russia, traditionally inhabited by the Sami people.⏎from WordNet 3.0 Copyright 2006 by Princeton University

  4. #4 peak

    which as explained here is only thought to be a possible explanation for dark matter in combination with Cold Dark Matter theory and not on its own.⏎This page was last updated June 27, 2015.<|endoftext|>05-1

  5. #5 peak

    , thin, or emaciated appearance; drawn; said of the face or the expression.⏎from WordNet 3.0 Copyright 2006 by Princeton University. All rights reserved.⏎- adj. somewhat ill or prone to illness

  6. #6 peak

    . FormsCan be used for anything<|endoftext|>We found 1,708 reviewed resources for state of the union address⏎How can word choice affect a political speech? Middle

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: 2, 25, 10, 14, 26, 11, 19, 12, 10, 10, 10, 10,

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

    6⏎- Saver Delivery: Yes⏎- 1st Class Delivery: Yes⏎- Courier Delivery: Yes⏎- Store Delivery: Yes⏎Prices are for internet purchases only. Prices and availability in WHSmith Stores may vary significantly⏎©

  2. #2 peak

    and directed by A. J. Hildebrand.)⏎Open content licensed under CC BY-NC-SA⏎Given an integer , the exponential sum version of the Gauss sum modulo is defined by⏎and the quadratic residue version is defined by⏎where is

  3. #3 peak

    our deepest concerns, judges us and guides us through our lives. A traditional Jewish concept of God is one that is interactive and intimately personal.⏎This week's Torah portion, Vaerah, begins with God hearing, and ultimately reacting to, the Israel

  4. #4 peak

    Saturday<|endoftext|>15.1: The Science Fair (5 minutes)⏎The purpose of this warm-up is for students to reason about two objects that have the same volume but different surface

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. L2:299 weakly excites
  2. L6:2939 weakly excites
  3. L4:182 weakly suppresses
  4. L8:774 weakly excites
  5. L5:706 weakly excites the number 1 starting a list
  6. L3:2117 weakly excites
  7. L5:1912 weakly excites
  8. L5:1588 weakly excites
  9. L7:3025 weakly suppresses
  10. L3:984 weakly excites "to" after a noun saying what to do
  11. L8:1221 weakly suppresses
  12. L0:2954 weakly suppresses
  13. L8:1474 weakly excites
  14. L0:413 weakly suppresses
  15. L5:164 weakly excites
  16. L8:224 weakly suppresses
  17. L2:1437 weakly suppresses
  18. L3:2915 weakly excites
  19. L3:1604 weakly excites
  20. L1:2967 weakly suppresses
  21. L1:1431 weakly excites
  22. L7:1438 weakly suppresses
  23. L1:2658 weakly suppresses
  24. L4:2256 weakly suppresses

Feeds later layers

  1. L10:1836 suppresses
  2. L17:1779 suppresses
  3. L14:1864 suppresses
  4. L14:1850 suppresses
  5. L17:2473 suppresses
  6. L18:1401 weakly suppresses
  7. L13:1162 weakly suppresses
  8. L13:2401 weakly suppresses
  9. L12:1378 weakly suppresses
  10. L22:1051 weakly suppresses
  11. L15:1909 weakly suppresses blank slot before a number
  12. L15:368 weakly suppresses
  13. L19:2743 weakly suppresses
  14. L14:1697 weakly suppresses
  15. L16:2069 weakly excites
  16. L15:1657 weakly suppresses
  17. L10:3023 weakly suppresses
  18. L13:2449 weakly suppresses
  19. L12:2168 weakly suppresses
  20. L15:2315 weakly suppresses
  21. L12:2811 weakly suppresses
  22. L23:1857 weakly suppresses
  23. L17:1488 weakly suppresses
  24. L16:3054 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.

1 values between -0.60 and -0.572 values between -0.54 and -0.511 values between -0.51 and -0.481 values between -0.48 and -0.452 values between -0.45 and -0.435 values between -0.43 and -0.404 values between -0.40 and -0.378 values between -0.37 and -0.348 values between -0.34 and -0.3118 values between -0.31 and -0.2828 values between -0.28 and -0.2533 values between -0.25 and -0.2256 values between -0.22 and -0.1979 values between -0.19 and -0.16128 values between -0.16 and -0.13202 values between -0.13 and -0.10276 values between -0.10 and -0.07399 values between -0.07 and -0.04540 values between -0.04 and -0.01584 values between -0.01 and 0.02477 values between 0.02 and 0.04313 values between 0.04 and 0.07220 values between 0.07 and 0.10160 values between 0.10 and 0.13119 values between 0.13 and 0.1676 values between 0.16 and 0.1962 values between 0.19 and 0.2225 values between 0.22 and 0.2528 values between 0.25 and 0.2817 values between 0.28 and 0.3114 values between 0.31 and 0.347 values between 0.34 and 0.377 values between 0.37 and 0.404 values between 0.40 and 0.433 values between 0.43 and 0.461 values between 0.54 and 0.57offnegativepositiveoften

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. 骅
  2. овый
  3. HandlerContext
  4. aca
  5. 的数据
  6. ólogo
  7. oci
  8. andan
  9. opport
  10. roup

Suppresses

  1. 저
  2. (nameof
  3. rounds
  4. (round
  5. 凌晨
  6. round
  7. ension
  8. 关门
  9. 一向
  10. ltd