textbook and study-help wording
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. 7 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 words common in textbooks, homework help and edition listings, such as 'answers', 'Edition', 'data' and 'expert'. When turned up, the model's text often shifted toward textbook problems and worked examples.
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.
-
ids and Bases practice, you can also practice Conjugate Acids and Bases practice problems.⏎What professor is this problem relevant for?⏎Based on our data, we think this problem is relevant for Professor Finnegan's class at WSU
-
just stop by for help. By getting the help of a tutor, the student can better understand their coursework.⏎Check the Textbook⏎Many textbooks will actually contain answers in the back of the book. For the more difficult problems, the
-
keeping us alive.⏎A good day for the amygd<|endoftext|>Calculate with Confidence | 5th Edition⏎AUTHOR: Gray Morris, Deborah C.⏎PUBLISHER: Else
-
th Edition⏎This is Herb's step-by-step introduction to Java, updated for⏎Java
-
Chegg Com⏎Derivation Of Kir<|endoftext|>The poisonous Gartraps are the bane
-
appropriate action is taken to protect the health of all Albertans.⏎You can protect yourself from all coronaviruses by:⏎- Cleaning your hands frequently using either soap and warm water or alcohol-based hand sanitizers,⏎- Covering your cough and
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 that, and if the following is true, then I can use the following and then I can have a certain degree of certainty, or 1, and 1, etc, and if that, and I can have the top,
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
lands and produce but can also punish the whole community for the act of an individual.⏎Into the bargain, the government has gone a step further and shifted the onus of proving innocence on the accused; the person is presumed to be guilty until they
-
least restrictive way for the government to further that interest.⏎Minor costs or inconveniences imposed on religious institutions are insufficient to trigger RLUIPA’s protections. The burden must be “substantial.” And, likewise, once the institution has shown a
-
2014|⏎|ISBN: ||9781312744967 (DRM-EPUB)|⏎
-
land to grow crops and raise livestock–“allowing less to be retained or restored as natural habitat” the<|endoftext|>Hardcover ISBN: 978150360048
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
- L3:1904 weakly excites
- L0:2015 weakly suppresses
- L5:46 weakly excites
- L3:2090 weakly excites
- L6:1291 weakly excites
- L5:28 weakly excites
- L1:2836 weakly excites
- L6:1476 weakly excites
- L6:1797 weakly suppresses
- L6:2885 weakly suppresses
- L5:180 weakly suppresses
- L6:2570 weakly suppresses
- L5:1495 weakly excites
- L5:94 weakly suppresses
- L3:1065 weakly excites
- L4:1745 weakly excites
- L6:106 weakly excites
- L1:2922 weakly excites
- L0:1333 weakly suppresses
- L6:1792 weakly suppresses
- L2:70 weakly excites
- L3:1545 weakly suppresses
- L3:2426 weakly excites
- L2:1334 weakly suppresses
Feeds later layers
- L15:1815 suppresses
- L8:1765 excites
- L10:1775 excites
- L9:2475 weakly excites
- L25:1250 weakly excites
- L22:2353 weakly excites the digit 0 in years 2000 to 2009
- L20:535 weakly suppresses
- L9:42 weakly excites
- L13:2463 weakly suppresses
- L8:1386 weakly excites
- L9:156 weakly suppresses
- L8:2995 weakly suppresses
- L8:1749 weakly suppresses
- L24:1594 weakly suppresses
- L8:1100 weakly excites
- L16:2752 weakly suppresses
- L11:2539 weakly suppresses
- L25:558 weakly suppresses
- L25:512 weakly suppresses
- L9:2141 weakly suppresses
- L27:2219 weakly suppresses
- L10:769 weakly suppresses
- L22:2182 weakly suppresses
- L11:1860 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
- 鲞
- ureka
- ứt
- ette
- 这些问题
- udas
- ruptcy
- busters
- 钯
- '](
Suppresses
- HomeController
- MainFrame
- enus
- fore
- sov
- 続けて
- -controls
- ctype
- ствовать
- conv