'is' or '1' after a variable
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. 8 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 fires on a short word or symbol that names or sits right after a variable, such as "A is" in a formula, "s1" in code, or "described" and "above" when pointing back to something said earlier. Its steering effects were unrelated to these cases, so the pattern is mixed.
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 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.
-
supercoiled, linear and open form of DNA?⏎The equation for calculating extinction coefficients is stated by Beer's Law:⏎A = epsilon.c.l⏎Where A is the absorbance measurement at a given wavelength⏎epsilon is the extinction coefficient in M
-
. It is one of over a hundred billion or so such stars inside our cosmic system, unremarkable aside from its status as our nearby neighbor. Circling Proxima is an intriguing yet puzzling world, called Proxmia b, found
-
. President Andrew Johnson. He was a colorful and often controversial figure on the national stage and in the Massachusetts political scene, during his one term as Governor.⏎Butler, a successful trial lawyer, served in the Massachusetts legislature as an antiwar Democrat
-
are two quanties that are in the same units of measure. The decibel is 0.1 bel, so the decibel value LdB is LdB = 10 log10 (P1/P0).⏎Define the mean
-
= 'Hello, this is Tom's laptop' >>> Output: >>> s1 = 'Hello, this is Tom's laptop' File "", line 1 s1 = 'Hello, this is Tom's laptop'⏎```The above code would give
-
,- 1) are orthogonal vectors in R 4 . Write the vector X = (1 , 1 , 1 , 2) in the form X = a U +<|endoftext|>Hepatitis B vaccine⏎The hepatitis B vaccine (
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: “If I was in a situation where I have to be a leader, but it’s not just a matter of getting more things to do, but being a bit more...” I think that maybe I just have to try to think of
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Only this way can the current and previous layers chemically bond and become a single solid object.⏎This phenomenon is known as overcuring. For the reasons just described, this overcuring is necessary and cannot be completely avoided.⏎However, over
-
using words effectively. The essay must be well researched and combine and analyze different sources and data. All these facilitate clear writing. An essay must have the qualities mentioned above. Additionally, it must be well-organized and formatted. The writing skills acquired
-
leads to reduced efficiency of the diesel engine itself, at the same time creates a remarkable potential for increased power output on both power turbines and steam turbines. The above mentioned potential compensates the reduction of diesel engine efficiency with surplus.⏎Accordingly, the
-
and alcohol consumption. Some people may snore every night, while others snore only occasionally.⏎What is sleep apnea?⏎Sleep apnea is similar to snoring, except that the airway becomes completely constricted. As a result, it
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
- L6:1040 excites
- L7:332 suppresses
- L7:352 excites
- L7:460 suppresses
- L7:269 excites
- L5:523 excites
- L5:904 excites
- L4:443 suppresses
- L6:484 excites
- L5:372 suppresses
- L6:116 excites
- L7:380 excites
- L7:1320 weakly excites
- L3:151 weakly suppresses
- L7:1521 weakly excites
- L6:277 weakly suppresses
- L7:99 weakly suppresses
- L7:1285 weakly suppresses
- L7:2514 weakly suppresses
- L7:1218 weakly excites
- L6:910 weakly suppresses
- L6:39 weakly suppresses
- L7:301 weakly suppresses
- L6:1102 weakly excites
Feeds later layers
- L10:1135 suppresses
- L10:581 suppresses
- L9:2458 suppresses
- L9:342 suppresses
- L12:590 excites
- L9:864 suppresses
- L11:2305 suppresses
- L11:1604 weakly suppresses
- L9:1368 weakly suppresses
- L13:2760 weakly suppresses
- L14:2912 weakly excites
- L16:3014 weakly excites
- L17:2094 weakly excites
- L14:1511 weakly suppresses
- L12:2515 weakly suppresses
- L15:1687 weakly excites
- L10:1973 weakly suppresses
- L14:2848 weakly suppresses
- L10:261 weakly suppresses
- L15:178 weakly suppresses
- L10:1735 weakly suppresses
- L23:1088 weakly excites
- L9:2738 weakly excites
- L11:2655 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
- )&&(
- стве
- .SwingConstants
- )||
- )||(
- สเต
- Onc
- avo
- ительно
- itur
Suppresses
- izen
- .places
- 床上
- 灰尘
- testData
- _bullet
- ��
- shale
- '//
- 愿