Neuron Explorer

Qwen3-0.6B

'to' after 'agreed'

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 fires on the word 'to' right after the verb 'agreed' or 'agrees', as in 'agreed to exchange' or 'agrees to receive'. Turning it up did not produce an obvious theme in the sample text.

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

    be used in the service of scientific research.⏎Article II: Establishes freedom of scientific research and encourages cooperation in the service of science.⏎Article III: Parties agree to exchange research and program plans, scientific results, and scientific personnel, in the spirit

  2. #2 peak

    trading over futures.⏎- Futures Market⏎The bitcoin futures market is similar to that of other assets. By purchasing a security with a futures contract, a buyer agrees to receive a security at a specified price and date. By buying a futures contract,

  3. #3 peak

    –2, 3–5, 6–8, 9–12⏎Important events during the presidency of Jimmy Carter.⏎President Jimmy Carter agreed to answer seven questions selected from several hundred posted by Scholastic students.⏎A time

  4. #4 peak

    , high-level U.S. officials assumed that a Warsaw Pact conventional or nuclear attack on NATO Europe would invite a U.S. nuclear response (unless the Soviets agreed to limit fighting to conventional weapons). Rejecting the idea of "no first use,"

  5. #5 peak

    second director). On March 3, 1879, the 45th Congress and President Hayes agreed to promote governmental economy and efficiency by discontinuing the three ongoing geological and geographical surveys and

  6. #6 peak

    points mean for adaptation support? A key recommendation is the need to learn from current local level processes of change – the changing risks<|endoftext|>On Friday Australian energy ministers agreed to an investigation into the country’s energy security, stability, and affordability to be led

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 following 199700000000000000000000000000000000000000000

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

    of the Romanov Dynasty⏎Role of Stalin in the 1917 Revolution⏎He was released from exile and became part of the editorial board of Pravda.⏎He was elected to the

  2. #2 peak

    is similar to use of arrowleaf balsamroot (pronghorn, bighorn sheep, and mule deer). Arrowleaf bals

  3. #3 peak

    bread; All power to the soviets.⏎He was caught up in a struggle in the Bolshevik party and called a betrayer of socialism because of the Pravda publishing support for the war effort and the provisional gouvernment.⏎He changed

  4. #4 peak

    soldiers to join the fight.⏎As the struggle against the Confederate South raged on, the newly-formed 54th Massachusetts battalion prepared to attack the seemingly impregnable fortress of Fort Wagner. The odds were heavily against them; the fort had

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. L0:2065 weakly excites
  2. L0:2976 weakly suppresses
  3. L0:2694 weakly suppresses
  4. L0:1655 weakly suppresses
  5. L0:2680 weakly suppresses
  6. L0:1040 weakly suppresses
  7. L0:2910 weakly suppresses
  8. L0:2151 weakly suppresses
  9. L0:2365 weakly excites mid-word pieces like 'advent' and 'precaution'
  10. L0:554 weakly suppresses
  11. L0:1301 weakly suppresses
  12. L0:1248 weakly suppresses
  13. L0:1379 weakly excites
  14. L0:807 weakly suppresses
  15. L0:1822 weakly suppresses
  16. L0:2541 weakly excites
  17. L0:2702 weakly excites
  18. L0:1025 weakly excites
  19. L0:2934 weakly suppresses
  20. L0:2284 weakly excites
  21. L0:1105 weakly suppresses
  22. L0:2390 weakly suppresses
  23. L0:1987 weakly suppresses
  24. L0:241 weakly suppresses

Feeds later layers

  1. L9:2365 excites
  2. L6:287 excites
  3. L14:1818 weakly excites
  4. L26:697 weakly suppresses
  5. L5:2742 weakly excites
  6. L20:722 weakly suppresses
  7. L15:2433 weakly suppresses
  8. L6:1133 weakly excites "to" before an infinitive verb
  9. L3:3013 weakly suppresses
  10. L19:1665 weakly suppresses
  11. L16:2304 weakly excites
  12. L25:2496 weakly suppresses
  13. L8:723 weakly excites
  14. L18:1790 weakly excites
  15. L2:1000 weakly suppresses
  16. L19:1930 weakly suppresses
  17. L13:2862 weakly excites
  18. L22:492 weakly suppresses
  19. L6:2310 weakly suppresses
  20. L5:1022 weakly suppresses
  21. L5:1582 weakly suppresses
  22. L22:268 weakly suppresses
  23. L13:2025 weakly excites
  24. L21:1928 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 -0.70 and -0.642 values between -0.64 and -0.571 values between -0.57 and -0.517 values between -0.51 and -0.445 values between -0.44 and -0.389 values between -0.38 and -0.3231 values between -0.32 and -0.2594 values between -0.25 and -0.19181 values between -0.19 and -0.13492 values between -0.13 and -0.061411 values between -0.06 and 0.001007 values between 0.00 and 0.07346 values between 0.07 and 0.13166 values between 0.13 and 0.1950 values between 0.19 and 0.2636 values between 0.26 and 0.3216 values between 0.32 and 0.3913 values between 0.39 and 0.457 values between 0.45 and 0.515 values between 0.51 and 0.585 values between 0.58 and 0.644 values between 0.64 and 0.703 values between 0.70 and 0.771 values between 0.77 and 0.834 values between 0.83 and 0.902 values between 0.90 and 0.963 values between 1.02 and 1.091 values between 1.15 and 1.221 values between 1.22 and 1.281 values between 1.28 and 1.341 values between 1.41 and 1.471 values between 1.79 and 1.85offnegativepositiveoften

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. oko
  2. обы
  3. long
  4. 绞
  5. azz
  6. /e
  7. ingt
  8. ulation
  9. ations
  10. thermal

Suppresses

  1. landers
  2. trailers
  3. lename
  4. adviser
  5. Territories
  6. hardest
  7. 录像
  8. 닝
  9. lander
  10. 纪录片