Neuron Explorer

Qwen3-0.6B

'real' as opposed to mock or simulated

A neuron in layer 7 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 'real' or 'realistic' when it is set against something fake, such as a mock MRI, real-life English or real-world interactions. Turned up, the model's writing moved toward real systems and real environments.

Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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

    prototyping shows people what the feature will look like. But the code base or project is not necessarily used for the production version of the application.⏎What is a mockup logo?⏎What Is A Logo

  2. #2 peak

    in school.⏎Finally, you may use the worksheets to construct and fortify your child’s standard math expertise. By using genuine-daily life example, worksheets can show how subtractions are performed, by

  3. #3 peak

    wooden frame covered with canvas and a large plastic tube. The mock MRI is completely safe and does not contain a magnet. Children who don’t want to enter a real MRI are invited to go into the mock scanner, where they are gradually exposed to

  4. #4 peak

    most effortless and effective learning happens while the student is having fun. The non-traditional environment of Life Club helps your child gain the skills and confidence to use real-life English.⏎As a component of the Efekta System, Life Club reinforces

  5. #5 peak

    ’s right, I said the “s” word.⏎When used right, animation is designed to mimic real-world interactions. In a way, we’ve come full-circle. We might not

  6. #6 peak

    that fits you best for each question below. If two fit...⏎Close enough to be relatable, yet distinct enough to show us a different option. Through a real life

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 value of a car and the temperature. This is a great place to talk about the specific model. The problem is the specific one, and the important part is 18.04. Then, the key is 4,

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

    emia II) More⏎(3) hydroxymethylglutaryl-CoA dehydrogenase deficiency; and (4) carnitine⏎The characteristic features of fatty acid oxidation disorders are mild⏎metabolic acid

  2. #2 peak

    , correcting hypoglycemia with⏎glucose, and providing oral carnitine at 100 mg/kg per day. Riboflavin⏎

  3. #3 peak

    difference between metabolism in cancer and normal cells is the switch in cancer to the production of a different version, or isoform, of a protein produced from the pyruvate kinase-M (PK-M) gene. The protein version produced in normal

  4. #4 peak

    the school. But ensure that these methods keep offenders in the educational system while also keeping other students and teachers safe.<|endoftext|>Blocking production of a pyruvate kinase splice-variant shows therapeutic promise⏎Cold Spring Harbor, N.Y.⏎– Cancer

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. L6:1549 weakly excites
  2. L6:1717 weakly suppresses
  3. L4:2181 weakly excites
  4. L5:1461 weakly suppresses
  5. L2:2851 weakly suppresses
  6. L1:2399 weakly excites
  7. L2:2133 weakly excites
  8. L2:2670 weakly suppresses
  9. L6:786 weakly excites
  10. L6:2856 weakly excites
  11. L6:2795 weakly excites
  12. L6:771 weakly excites
  13. L3:2039 weakly suppresses
  14. L2:1982 weakly excites
  15. L2:2807 weakly excites
  16. L5:961 weakly suppresses 'area' and 'volume' as measures
  17. L6:389 weakly suppresses
  18. L2:1477 weakly excites
  19. L0:2654 weakly excites
  20. L4:2589 weakly excites
  21. L0:2337 weakly suppresses
  22. L3:1956 weakly excites
  23. L4:537 weakly excites
  24. L3:115 weakly suppresses

Feeds later layers

  1. L10:2748 excites
  2. L17:2656 excites
  3. L25:2425 excites
  4. L22:1416 excites
  5. L12:883 weakly excites
  6. L8:838 weakly excites
  7. L14:1459 weakly excites
  8. L12:1272 weakly excites
  9. L20:1920 weakly excites
  10. L14:1773 weakly excites
  11. L15:1161 weakly excites
  12. L10:2053 weakly suppresses
  13. L11:1162 weakly excites
  14. L9:2818 weakly suppresses
  15. L19:1960 weakly excites
  16. L13:2415 weakly suppresses
  17. L8:2600 weakly excites
  18. L8:470 weakly suppresses
  19. L9:337 weakly excites
  20. L9:1762 weakly suppresses
  21. L8:2534 weakly excites
  22. L8:2852 weakly excites
  23. L10:2589 weakly excites
  24. L24:2443 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 -2.45 and -2.281 values between -2.28 and -2.111 values between -1.94 and -1.762 values between -0.91 and -0.736 values between -0.73 and -0.5626 values between -0.56 and -0.39209 values between -0.39 and -0.221201 values between -0.22 and -0.051979 values between -0.05 and 0.12393 values between 0.12 and 0.3054 values between 0.30 and 0.4715 values between 0.47 and 0.644 values between 0.64 and 0.815 values between 0.81 and 0.983 values between 0.98 and 1.152 values between 1.15 and 1.321 values between 1.67 and 1.841 values between 1.84 and 2.011 values between 2.52 and 2.701 values between 2.87 and 3.041 values between 4.24 and 4.41offnegativepositiveoften

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. real
  2. real
  3. 真实
  4. 真实的
  5. 实际
  6. 实
  7. (real
  8. REAL
  9. 真實
  10. -real

Suppresses

  1. NewLabel
  2. )section
  3. 楣
  4. 亢
  5. slogan
  6. cest
  7. ergus
  8. Purs
  9. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
  10. ))==