Neuron Explorer

Qwen3-0.6B

'it' before 'is important to'

A neuron in layer 23 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

The neuron lights up on the word 'it' when it starts a sentence after a phrase like 'Therefore,' and is followed by a phrase such as 'is important to' or 'is required'. When it was turned up, the model got stuck repeating the word 'is'.

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

    will have a significant impact on your health and fitness. Therefore, it is important to choose a particular school of yoga. Ashtanga yoga can be

  2. #2 peak

    textile materials in the single light source will not be adequate. To avoid such circumstances, it is required to test the colours from the fabrics in distinctive light shades with the goal

  3. #3 peak

    the dangerous gases owing to the following properties:⏎Tips for Safe Storage of Compressed Acetylene Gas Cylinders⏎Owing to the above discussed hazards, it has become important to take special care while storing and handling compressed gas cylinders of ac

  4. #4 peak

    essential to ensure that it receives sufficient stimulus for muscle growth. Not all muscles working in a multi-joint movement will necessarily produce the same muscle growth. Therefore, it may be advisable to include additional exercises that specifically target the muscles with stimulating repetitions in

  5. #5 peak

    the classification of cloud computing contracts in the Turkish and European legal frameworks.⏎1. Cloud computing and cloud computing service models⏎Prior to classifying cloud computing contracts, it is important to understand what kind of services are provided pursuant to such contracts and who

  6. #6 peak

    symptoms depends on many factors, including your dog's size and how much of the plant he ingested. Severe cases can result in death, however, so it's vital to contact your veterinarian immediately if you suspect your dog has chewed on

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: should one should not need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to need to

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

    the candidate pool more quickly.⏎If every question halved the size of the candidate pool, then four or five closed (yes/no) questions should be all that’s required. However, Guess Who is

  2. #2 peak

    found all the ways?⏎Swap the stars with the moons, using only knights' moves (as on a⏎chess board). What is the smallest number of moves possible?⏎Can you work out how to balance this equaliser? You can put

  3. #3 peak

    can it recover low-grade heat for power generation, it can also be done using thinner, lighter devices than semi-conductors. One slim cell is all that is needed to regulate the thermal cycle, with graphic oxide applied on one side

  4. #4 peak

    to look for:⏎No special preparation is needed before the test. A chest x-ray will be performed before and after the

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. L22:912 strongly excites
  2. L22:1041 suppresses
  3. L20:2349 excites
  4. L20:2994 excites
  5. L19:2765 excites
  6. L21:2956 suppresses
  7. L22:1588 excites
  8. L20:2182 suppresses
  9. L21:1990 excites
  10. L22:691 suppresses
  11. L19:1597 suppresses
  12. L21:1827 suppresses
  13. L20:2706 suppresses
  14. L18:1813 suppresses
  15. L18:2998 excites
  16. L22:1227 excites
  17. L17:246 excites
  18. L21:1199 excites
  19. L15:2584 suppresses
  20. L20:2818 suppresses
  21. L20:1696 weakly excites
  22. L21:2280 weakly excites
  23. L18:522 weakly suppresses
  24. L14:2332 weakly suppresses

Feeds later layers

  1. L26:109 strongly suppresses
  2. L27:500 excites
  3. L25:822 excites
  4. L27:804 excites
  5. L24:565 excites
  6. L24:299 excites
  7. L24:319 excites
  8. L26:1182 excites
  9. L26:978 weakly excites
  10. L24:1279 weakly excites
  11. L27:1256 weakly excites
  12. L27:1236 weakly suppresses
  13. L24:789 weakly suppresses
  14. L24:764 weakly suppresses
  15. L26:1276 weakly suppresses
  16. L25:1674 weakly excites
  17. L24:1690 weakly excites
  18. L24:2503 weakly excites
  19. L24:240 weakly suppresses
  20. L27:1123 weakly suppresses
  21. L25:9 weakly suppresses
  22. L25:1471 weakly suppresses
  23. L24:1584 weakly suppresses
  24. L27:1275 weakly 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 -11.83 and -10.901 values between -10.90 and -9.981 values between -9.98 and -9.051 values between -9.05 and -8.131 values between -6.28 and -5.356 values between -5.35 and -4.436 values between -4.43 and -3.508 values between -3.50 and -2.579 values between -2.57 and -1.65260 values between -1.65 and -0.722784 values between -0.72 and 0.20750 values between 0.20 and 1.1342 values between 1.13 and 2.0515 values between 2.05 and 2.9811 values between 2.98 and 3.903 values between 3.90 and 4.831 values between 4.83 and 5.752 values between 5.75 and 6.682 values between 6.68 and 7.611 values between 7.61 and 8.531 values between 9.46 and 10.381 values between 12.23 and 13.161 values between 24.26 and 25.19offnegativepositiveoften

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. is
  2. 是
  3. .is
  4. 是比较
  5. 是非常
  6. 是很
  7. Is
  8. is
  9. là
  10. 的是

Suppresses

  1. must
  2. 必须
  3. moeten
  4. ENCED
  5. deben
  6. 应该
  7. 应在
  8. otron
  9. accountId
  10. 应