Neuron Explorer

Qwen3-0.6B

5' notation and DNA biology

A neuron in layer 6 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featured, so it has steering outputs.

What this neuron does

This neuron mostly responds to the prime mark in '5'' when it refers to the ends of DNA or RNA, and to DNA repair and nuclease terms like 'endonuclease'. It tends to go quiet on everyday product words such as 'memory foam' and 'shock absorption'. Turning it up did not bring out a clear theme.

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

    . The miRNA is used as a template to synthesize cDNA using oligo-dT and Reverse Transcriptase. A universal tag sequence on the 5' end of oligo-dT primers facilitates the amplification of cDNA in

  2. #2 peak

    for consensus intron/exon sequences:⏎Analysis of a large number of mRNA genes has led to the identification of highly conserved consensus sequences⏎at the 5' and 3' ends of essentially all mRNA introns.(Figure⏎of Cons

  3. #3 peak

    homology dependent DSB repair (HR), non-homologous end joining (NHEJ), the Fanconi Anemia pathway (FANC

  4. #4 peak

    even more tightly to the endonuclease active site and thus will potentially be more potent inhibitors of influenza virus replication.⏎

  5. #5 peak

    the Ataluren pontent inhibitor pathways/proteins that are activated in response to each type of damage. A. A double strand break DNA break (DSB) most often results in the activation, by phosphorylation (indicated in this

  6. #6 peak

    in France focused on what is known as the endonuclease part of the enzyme RNA polymerase.⏎The endonuclease enables for the virus to trick someone into producing viral proteins, which join into

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: a 1000 N/m system. How can I find the natural frequency of the system, and what is the damping coefficient for the mass? The system is described by a mass-spring system with a damping force, and 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

    mainly in the axial direction.⏎Are car bumpers made of plastic?⏎The vast majority of modern bumpers are made of plastic, a material that provides effective shock absorption while minimizing the vehicle’s weight and cost.⏎What is behind a bumper cover?⏎

  2. #2 peak

    night’s sleep. It also performs marvels of engineering.⏎Memory foam was first invented in the late 1960s and ’70

  3. #3 peak

    material would be an excellent fit for the mattress industry. The first Tempur-Pedic® mattress came out in 1991, and soon, memory foam enjoyed a huge surge in popularity.⏎How Memory Foam Is Made⏎There are six

  4. #4 peak

    9.⏎Since 1923, the IIP has annually dropped two to three memorial<|endoftext|>Introduction. In automotive industry, bumper beams are used as shock absorbing parts. They are attached to the front and rear ends of motor vehicles

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. L3:1048 suppresses
  2. L5:75 weakly excites
  3. L5:347 weakly suppresses
  4. L5:1399 weakly excites
  5. L5:2142 weakly suppresses
  6. L5:197 weakly excites
  7. L3:1387 weakly excites
  8. L1:2801 weakly excites
  9. L5:46 weakly excites
  10. L5:50 weakly excites
  11. L0:1301 weakly suppresses
  12. L5:222 weakly suppresses
  13. L5:839 weakly suppresses
  14. L5:1081 weakly suppresses
  15. L3:322 weakly excites
  16. L0:1033 weakly suppresses
  17. L4:1427 weakly suppresses
  18. L5:2553 weakly excites
  19. L4:214 weakly suppresses
  20. L1:3050 weakly suppresses
  21. L0:2779 weakly suppresses
  22. L5:750 weakly suppresses
  23. L0:1224 weakly suppresses
  24. L0:1061 weakly suppresses

Feeds later layers

  1. L7:1861 suppresses
  2. L7:2997 suppresses
  3. L8:2846 suppresses
  4. L8:1245 suppresses
  5. L7:2383 suppresses
  6. L22:1754 weakly suppresses
  7. L8:2125 weakly suppresses
  8. L8:1957 weakly suppresses
  9. L10:780 weakly suppresses
  10. L20:2245 weakly excites
  11. L10:2478 weakly suppresses
  12. L20:447 weakly suppresses
  13. L10:2239 weakly suppresses
  14. L14:1824 weakly suppresses
  15. L11:1576 weakly excites
  16. L10:2886 weakly excites
  17. L21:2702 weakly excites
  18. L7:1731 weakly suppresses
  19. L8:2358 weakly excites
  20. L9:2613 weakly suppresses
  21. L9:1097 weakly excites
  22. L15:2543 weakly suppresses
  23. L9:3042 weakly suppresses
  24. L11:812 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 -3.26 and -3.131 values between -1.22 and -1.091 values between -0.83 and -0.713 values between -0.71 and -0.589 values between -0.58 and -0.4554 values between -0.45 and -0.33272 values between -0.33 and -0.20901 values between -0.20 and -0.071608 values between -0.07 and 0.06793 values between 0.06 and 0.19201 values between 0.19 and 0.3143 values between 0.31 and 0.448 values between 0.44 and 0.572 values between 0.57 and 0.702 values between 0.70 and 0.822 values between 0.82 and 0.953 values between 0.95 and 1.081 values between 1.21 and 1.332 values between 1.33 and 1.461 values between 1.71 and 1.84offnegativepositiveoften

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. 莓
  2. ursor
  3. раг
  4. luent
  5. rogen
  6. opus
  7. 注
  8. ocratic
  9. 菌
  10. issippi

Suppresses

  1. yı
  2. Ny
  3. mov
  4. 敝
  5. ControllerBase
  6. asmus
  7. achts
  8. Hear
  9. deaf
  10. wię