Neuron Explorer

Qwen3-0.6B

unclear: commas and nouns before an aside

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 is hard to pin down. It gives only weak signals, mostly on commas or nouns that come just before an extra clarifying phrase. It also turns negative on nouns like 'spine' or 'vessel'. Steering showed no clear pattern.

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

    apply to your risk is helpful, risks must also be analyzed in relation to each other. That’s why developing a scoring system that works for your organization’s risk, threat, and vulnerability inventory is important to building a tailored

  2. #2 peak

    man to do just that. After discussion with another friend, Wes Prichard, Prichard suggested to Wennberg<|endoftext|>Home > Preview⏎The flashcards below were created by user⏎on FreezingBlue Flashcards.⏎What factors could affect

  3. #3 peak

    to engage!⏎***Just a friendly 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,

  4. #4 peak

    the mind, according to a new paper that argues the disease is a sort of cousin of diabetes.⏎But this theory of anorexia as a fundamentally biological disorder, rather than a psychological one, is untested, psychiatrists warn, and patients

  5. #5 peak

    human genetics counselor will play an increasingly important role as patients and doctors struggle with the impact of these exciting new breakthroughs. The Human Genetics Program at Sarah Lawrence College the oldest and largest training program for genetic counselors in the United States, is well positioned

  6. #6 peak

    Henry III himself reissued Magna Carta a third time, this time in a shorter version with only <|endoftext|>ATHENS – The digital exhibition Proving History, presented by the Hellenic Ministry of

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: what is the number of integer solutions to the problem of the problem of the problem of the problem of the problem of the problem of the problem of the problem of the problem of the problem of the problem of the problem of the problem of the

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

    in particular your transverse abdoinimis (TA), you have to pull your belly button in toward your spine. Pulling your navel toward your spine is<|endoftext|>CMS Web Guide (University of Michigan-Flint)⏎What is a

  2. #2 peak

    cause frequent congestion, which can irritate the membranes and blood vessels of the nose. Medications that are used to treat conditions like sinusitis, colds and allergies (such as decongestants and histamines) can also dry up the

  3. #3 peak

    ots.⏎What Is Thrombosis?⏎Thrombosis is when the body forms a blood clot inside a blood vessel. A blood clot in a blood vessel is also called a thrombus. A thrombus can stop blood flow to parts

  4. #4 peak

    they don’t look like those letters—at best they would be Procrustean fits. For example, the Great Depression of 1929-33 saw economic activity slide steeply in 1929

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. L13:2609 strongly suppresses
  2. L13:1232 excites
  3. L12:1792 suppresses
  4. L11:1579 excites
  5. L12:616 excites
  6. L8:511 excites
  7. L11:782 excites
  8. L11:237 suppresses
  9. L13:706 excites
  10. L7:419 excites
  11. L5:694 excites
  12. L9:749 suppresses
  13. L5:99 excites
  14. L8:668 suppresses
  15. L2:180 suppresses 'is' after focus or goal nouns
  16. L13:1014 suppresses
  17. L11:344 excites
  18. L13:2835 excites
  19. L11:578 suppresses
  20. L9:542 excites
  21. L9:356 excites
  22. L13:2983 suppresses
  23. L9:1579 suppresses
  24. L6:250 excites

Feeds later layers

  1. L15:1543 suppresses
  2. L15:1176 suppresses
  3. L16:559 suppresses weak, mixed: verbs and adverbs in claims
  4. L16:1315 suppresses word or phrase being named or defined
  5. L17:1065 suppresses
  6. L15:233 suppresses
  7. L17:530 suppresses
  8. L16:749 suppresses
  9. L15:1513 suppresses
  10. L15:2554 suppresses
  11. L16:295 excites
  12. L18:391 suppresses
  13. L15:64 suppresses
  14. L15:423 suppresses
  15. L18:933 suppresses
  16. L15:1571 excites
  17. L16:1041 suppresses
  18. L15:161 suppresses
  19. L16:1886 excites
  20. L15:2234 excites
  21. L15:594 weakly suppresses
  22. L15:1236 weakly excites
  23. L16:918 weakly suppresses
  24. L15:827 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 -4.77 and -4.621 values between -4.17 and -4.021 values between -4.02 and -3.871 values between -3.58 and -3.432 values between -2.98 and -2.831 values between -2.83 and -2.681 values between -2.68 and -2.531 values between -2.53 and -2.381 values between -2.23 and -2.085 values between -2.08 and -1.934 values between -1.93 and -1.795 values between -1.79 and -1.641 values between -1.64 and -1.494 values between -1.49 and -1.342 values between -1.34 and -1.196 values between -1.19 and -1.045 values between -1.04 and -0.8910 values between -0.89 and -0.744 values between -0.74 and -0.598 values between -0.59 and -0.4420 values between -0.44 and -0.2993 values between -0.29 and -0.14674 values between -0.14 and 0.011813 values between 0.01 and 0.16588 values between 0.16 and 0.30262 values between 0.30 and 0.45152 values between 0.45 and 0.60107 values between 0.60 and 0.7568 values between 0.75 and 0.9046 values between 0.90 and 1.0520 values between 1.05 and 1.20offnegativepositiveoften

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. `:
  2. Pivot
  3. Rock
  4. ':
  5. 抽出
  6. ashed
  7. 一味
  8. 方方面
  9. adj
  10. **:

Suppresses

  1. ulet
  2. кова
  3. edReader
  4. heimer
  5. inerary
  6. loff
  7. awi
  8. 副县长
  9. assertInstanceOf
  10. enstein