Neuron Explorer

Qwen3-0.6B

weak end-of-document signal, mostly noise

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 responds only weakly, mostly to the marker that ends one document and starts another, and it is pushed slightly down on some words inside sentences. Its strongest responses are small, so it is hard to read. Turning it up did not produce a clear theme.

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

    of the medical staff at Phoenixville Hospital. His office is located at 824 Main St., Suite 206.<|endoftext|>MEDIA RELEASE - "CLIMATE IMPACTS DAY" BASED ON MISUNDER

  2. #2 peak

    general radiology that can be executed through the use of X-rays, mammography, ultrasound, fluoroscopy, angiography, CT scan, PET scan,<|endoftext|>Geophysical research in Thailand are relatively new due to the lack of fund to obtain

  3. #3 peak

    ifolds and functions of many complex variables. Application of transcendental methods to algebraic geometry falls in this category, together with more geometric chapters of complex analysis.<|endoftext|>By: Ann Gabriel| Foodfindsasia.com⏎As Filipinos, we

  4. #4 peak

    mutations arise over time⏎An NIH-supported team offers a possible answer to this question in a study published in the journal Science. Based on their analysis of duplicate gene<|endoftext|>The maternal mortality rate reported in Tibet (400/100,

  5. #5 peak

    the month.⏎VSTD9 Oscilloscope Operation & Signal Test Methods⏎Use our online self assessment tool to find the right course for you⏎Using our '<|endoftext|>Dividing by zero. Even on our own wiki, at one point.⏎Hand

  6. #6 peak

    this problem web Miner's has introduced several preprocessing methods to reduce the log data size and to eliminate incorrect records, which lead to improved quality, quantity and performance.<|endoftext|>Based on the acid rain and concurrent meteorological observational data during the past 1

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 question is, why do we think that the amount of time we spend on the topic is the same as the amount of time we spend on the topic, and then again again, the same amount of time, and again again, 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

    After the Patagonia incident, many other brands underwent investigation and uncovered that their own down also came from similarly unethical sources.⏎According to PETA, plucking live birds causes “considerable pain and distress”. When birds are plucked, their

  2. #2 peak

    Europe from the 4th to 6th centuries AD, when the Western Roman Empire was in decline.”⏎Pictish communities…were part of powerful kingdoms in the early medieval period.Dr Gordon Noble, Senior Lecturer, University of Aberdeen

  3. #3 peak

    health effects of consuming GMOs/Roundup shown in animal feeding studies⏎There’s now overwhelming evidence that GMOs are so dangerous and high risk that they should never have been introduced into the food supply. You won’t hear that statement by companies

  4. #4 peak

    ary [Video File]. Retrieved from https://youtu.be/qlSxPouPCIM⏎The post 1929 Stock Market Crash and the Great Depression. first appeared on COMPLIANT PAPERS.⏎The post

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. L11:695 suppresses
  2. L12:1190 excites
  3. L11:990 suppresses
  4. L11:710 excites
  5. L12:830 excites
  6. L11:893 excites
  7. L12:805 suppresses
  8. L11:1348 suppresses
  9. L12:741 suppresses
  10. L12:800 excites
  11. L12:1307 excites
  12. L11:931 suppresses
  13. L12:396 suppresses
  14. L9:99 suppresses
  15. L12:2488 suppresses
  16. L11:1308 excites
  17. L9:137 excites
  18. L8:82 suppresses
  19. L12:1026 suppresses
  20. L11:277 suppresses
  21. L7:185 excites
  22. L12:1073 excites
  23. L9:305 suppresses
  24. L12:2270 suppresses

Feeds later layers

  1. L16:491 strongly suppresses
  2. L14:670 excites
  3. L14:697 excites
  4. L16:472 suppresses
  5. L14:1379 excites
  6. L15:341 excites
  7. L14:384 excites
  8. L14:108 suppresses
  9. L15:2381 excites
  10. L14:1617 excites
  11. L15:1392 excites
  12. L14:2277 excites
  13. L14:2790 excites
  14. L16:2717 excites
  15. L17:2264 excites
  16. L16:2806 excites
  17. L17:511 excites
  18. L14:185 excites
  19. L16:2919 excites
  20. L15:567 excites
  21. L16:343 excites
  22. L16:433 excites
  23. L17:843 suppresses
  24. L14:1294 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 -2.46 and -2.341 values between -2.09 and -1.971 values between -1.97 and -1.851 values between -1.85 and -1.731 values between -1.61 and -1.492 values between -1.49 and -1.371 values between -1.37 and -1.254 values between -1.25 and -1.136 values between -1.13 and -1.005 values between -1.00 and -0.8814 values between -0.88 and -0.7628 values between -0.76 and -0.6433 values between -0.64 and -0.5260 values between -0.52 and -0.40143 values between -0.40 and -0.28283 values between -0.28 and -0.16799 values between -0.16 and -0.041665 values between -0.04 and 0.09572 values between 0.09 and 0.21212 values between 0.21 and 0.3351 values between 0.33 and 0.4512 values between 0.45 and 0.572 values between 0.57 and 0.692 values between 0.69 and 0.811 values between 0.93 and 1.051 values between 1.05 and 1.172 values between 1.17 and 1.301 values between 1.30 and 1.421 values between 1.42 and 1.541 values between 1.54 and 1.661 values between 1.66 and 1.781 values between 2.26 and 2.39offnegativepositiveoften

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. locations
  2. uely
  3. nationals
  4. enan
  5. acula
  6. location
  7. ulous
  8. anning
  9. phasis
  10. 璨

Suppresses

  1. .Env
  2. ..<
  3. 生态圈
  4.   
  5. тика
  6. th
  7. до
  8. _sy
  9. ViewItem
  10. NSE