Neuron Explorer

Qwen3-0.6B

'for' right after 'look'

A neuron in layer 1 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 lights up on the word 'for' when it comes straight after 'look', as in 'look for clinics'. It is a very strong, consistent trigger, but turning it up did not make the model's writing about searching or looking.

Written by Claude from this neuron's activations and steering outputs. Confidence: medium. Think it is wrong? Flag it (opens an email to the maker with the neuron already filled in).

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 listing of the many different types of clinics that do molecular genetic screening at http://cagc-accg.ca (click on Genetic Clinics). Look for clinics that do ocular genetics, general genetics or paediatric genetics - not cancer

  2. #2 peak

    (Engine Off)⏎- Look for puddles on the ground.⏎- Look for dripping fluids on underside of engine and

  3. #3 peak

    , your baby will show pleasure at the sight and sound of their parents, and may get sad when they don't see them.⏎- Step 6: Look for cause and effect recognition at 3 months, as the child picks up and drops

  4. #4 peak

    to take two or more sample days or weeks apart and show that the RSV antibodies have increased). We can also use antigen-antibody tests to look for the virus: these tests give

  5. #5 peak

    plwalsh: I'm glad it wasn't just us who saw the poor editing.”⏎‘golden-eye’ hasn't been added to any lists yet.⏎Looking for tweets for golden-eye.<|endoftext|>New subjects guide for primary years⏎Designated

  6. #6 peak

    tool that can identify approximately 8,500 distinct groups of bacteria in a single assay, the scientists profiled the organisms present in each sample to look for relationships between bacterial community composition and clinical characteristics of the patients' asthma.The researchers found

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 been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately. I have been thinking about lately

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

    that there is really no way to avoid jargon. It is present in every (I mean every) activity and profession and interest in the entire world from skydiving to sewing. Every human activity has its own special set of insider terms.⏎

  2. #2 peak

    -- hospitals, clinics, private practices, nursing homes, colleges, industry, and the armed forces.⏎Check with people you know who have been helped by a podiatrist. Other healthcare providers

  3. #3 peak

    Masters in Liberal Studies Program, and various master's programs including psychology, journalism, and biochemistry to name a few as well as some continuing education requirements. Check with your institution and program.⏎COURSE DESCRIPTION: An exploration of biological theories of human

  4. #4 peak

    and a variety of skin issues. Dramamine or Bonine are also useful for dogs suffering from motion sickness. Before giving your dog an OTC medication, check with your vet, since they could worsen certain conditions or interact with other medicines.⏎

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. L0:2363 weakly suppresses
  2. L0:3039 weakly suppresses
  3. L0:2906 weakly suppresses
  4. L0:217 weakly suppresses
  5. L0:2620 weakly excites
  6. L0:1931 weakly suppresses
  7. L0:1360 weakly excites
  8. L0:1301 weakly suppresses
  9. L0:2936 weakly excites
  10. L0:178 weakly suppresses
  11. L0:2845 weakly suppresses
  12. L0:2568 weakly suppresses
  13. L0:1235 weakly excites
  14. L0:1849 weakly suppresses
  15. L0:2340 weakly suppresses
  16. L0:2977 weakly suppresses
  17. L0:873 weakly suppresses
  18. L0:3046 weakly excites
  19. L0:1424 weakly excites
  20. L0:1434 weakly suppresses
  21. L0:2761 weakly excites
  22. L0:1388 weakly excites
  23. L0:1542 weakly excites
  24. L0:1330 weakly excites

Feeds later layers

  1. L2:1364 weakly excites
  2. L15:49 weakly suppresses
  3. L16:1671 weakly suppresses
  4. L23:850 weakly excites
  5. L8:2842 weakly suppresses
  6. L11:2645 weakly excites
  7. L19:2406 weakly excites
  8. L23:2629 weakly excites
  9. L3:1662 weakly suppresses
  10. L7:549 weakly suppresses
  11. L21:2936 weakly excites
  12. L20:2274 weakly excites
  13. L19:2809 weakly excites comma after 'in this blog post'
  14. L4:1682 weakly suppresses
  15. L14:1393 weakly excites
  16. L11:1265 weakly suppresses
  17. L10:2384 weakly excites place and journal names in citations
  18. L26:2919 weakly excites
  19. L20:1171 weakly excites
  20. L10:886 weakly excites
  21. L17:226 weakly suppresses
  22. L11:1116 weakly excites
  23. L5:2447 weakly suppresses
  24. L22:2629 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 -1.53 and -1.321 values between -1.32 and -1.112 values between -1.11 and -0.8913 values between -0.89 and -0.6813 values between -0.68 and -0.4735 values between -0.47 and -0.26879 values between -0.26 and -0.052812 values between -0.05 and 0.16114 values between 0.16 and 0.3713 values between 0.37 and 0.587 values between 0.58 and 0.795 values between 0.79 and 1.003 values between 1.00 and 1.213 values between 1.21 and 1.422 values between 1.42 and 1.632 values between 1.63 and 1.841 values between 1.84 and 2.051 values between 2.89 and 3.101 values between 6.68 and 6.89offnegativepositiveoften

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. lights
  2. 嵘
  3. Patterns
  4. odel
  5. pattern
  6. .Empty
  7. ifter
  8. ippets
  9. iore
  10. руд

Suppresses

  1. 醋
  2. ирование
  3. 发动
  4. ulture
  5. 舐
  6. }()⏎
  7. 直播
  8. filme
  9. ecast
  10. 直通车