unclear: physics constants and formulas
A neuronOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in layerOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. 20 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. It is featuredA neuron picked for the full treatment: steering outputs and a written explanation. About 3,000 of the 86,000 neurons are featured., so it has steering outputs.
What this neuron does
This neuron responds to scientific facts, such as the speed of light, heat radiation formulas and the Stefan-Boltzmann law. The pattern is loose, and turning it up did not clearly make the model write about physics.
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 activationThe number a neuron produces for a token. Large positive means it fired hard; negative means it pushed the other way. was highest. Amber marks positive activation; the outlined tokenA piece of text the model reads at once: often a word, sometimes part of a word or a punctuation mark. is the peak. Hover a token for its strength.
-
temperature)⏎and E is the flux in W/m²⏎More about this formula and background on the material properties in Planck, Stefan-Boltzmann, Kirchhoff and LTE.⏎The parameter called emissivity is the focus of
-
water is said to have a greater optical density than air.⏎Substance Speed of light (m/sec)⏎Vacuum 3 x 108⏎Air 3 x 108⏎Water 2.258 x
-
various people concerned that no one has really been interested in measuring surface emissivity properly.⏎All solid and liquid surfaces emit thermal radiation according to the Stefan-Boltzmann formula:⏎E = εσT4⏎where ε=emissivity,
-
be approximated by:⏎J = D (CWaterlooEmitter - groundwater) / membrane wall thickness⏎where D is the diffusion coefficient for the membrane material and is expressed in L2 / t.⏎Note: In gas release applications, mass is
-
of light changes and bending of light occurs.⏎In air, light travels at 3 x 108 m/sec but in water, it travels only about 2.258 x 108 m
-
Sunday, March 26, in Chase Hall Lounge, Bates College. The public is invited to attend without charge.⏎Light normally travels through a vacuum at about 186,000 miles per second. Hau and her team
SteeringAdding a push along a neuron while the model writes, to see how its output changes. Turning it up or down.
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.
Here is something I have been thinking about lately: I have a 4x3x3x3xxtogframe (the standard 3D version of the 4-ogtrion), and I have a trick in the right behind the 3ogtrion. I
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
able to properly support your body; allow you to stand, walk, run, and jump; and absorb damaging shock that enters your body every time your heel hits the ground. The following are interesting facts about your feet:⏎What Causes Plantar Fasc
-
the recording below.<|endoftext|>Normal Anatomy of the Knee Joint⏎The knee is made up of four bones. The femur or thighbone is the bone connecting the hip to the knee. The tibia or shinbone connects the knee to the
-
Email to a Friend|⏎Pleural fluid analysis is a test that examines a sample of fluid that has collected in the pleural space. This is the space between the lining of the outside of the lungs (pleura) and the wall of
-
related to the number and type of joints). An extreme case with especially high degrees of freedom is given by systems in which muscles are unattached to any internal or external skeleton whatsoever. Consequently, these structures are much more flexible than jointed limbs and
Connections
Show on mapWhich neuronsOne of thousands of tiny dials inside each layer. It reads the text so far and produces a number, its activation, that nudges what the model writes next. in earlier layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.
Fed by earlier layers
- L14:717 excites
- L10:876 suppresses
- L5:132 excites
- L6:2167 excites
- L7:87 excites
- L4:1421 excites
- L3:1048 suppresses
- L9:2087 suppresses
- L3:784 suppresses
- L3:388 suppresses
- L10:291 suppresses
- L9:628 suppresses
- L9:2792 suppresses
- L9:2934 weakly excites
- L8:1814 weakly suppresses
- L1:366 weakly excites
- L6:1254 weakly excites
- L9:473 weakly excites
- L8:2125 weakly excites
- L7:91 weakly excites
- L4:905 weakly excites
- L9:411 weakly suppresses
- L8:2846 weakly excites
- L10:68 weakly suppresses
Feeds later layers
- L23:2044 weakly excites
- L23:1471 weakly suppresses
- L24:2162 weakly suppresses
- L22:2932 weakly suppresses
- L23:1946 weakly excites
- L26:1412 weakly suppresses
- L22:2344 weakly suppresses number after 'account for' or 'over'
- L21:2092 weakly suppresses
- L22:1451 weakly suppresses
- L24:2487 weakly suppresses
- L26:1881 weakly suppresses
- L21:2058 weakly suppresses
- L23:2224 weakly suppresses
- L22:634 weakly excites
- L21:2082 weakly suppresses
- L24:1840 weakly suppresses
- L24:2291 weakly suppresses
- L24:1164 weakly suppresses
- L22:2456 weakly suppresses
- L23:1686 weakly suppresses
- L24:729 weakly excites
- L22:1960 weakly suppresses
- L21:2674 weakly suppresses
- L24:2676 weakly suppresses
Activation histogramA chart of how often a neuron takes each activation value.
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.
Logit lensReading which output tokens a neuron pushes toward or away from, straight off its output weights, before the later layers have their say.
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 layersOne step in the stack the text passes through. Qwen3-0.6B has 28 layers, each with 3,072 neurons. instead.
Promotes
- relationship
- relationships
- 关系
- law
- law
- 申
- declines
- 旻
- Law
- _relationship
Suppresses
- vergence
- ceği
- idia
- ertility
- .createUser
- .Unmarshal
- xlim
- ucken
- mutate
- gregator