"(" before a year in citations
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 very consistent: they mostly peak on the same token. 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
Fires on the opening bracket in reference lists, right before the year of a citation, such as "Kunkel, A. (2010)". Turning it down or up did not produce any clear theme in the writing.
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.
-
. 37–47)⏎Preston, J. D., O’Neal, J. H., & Talaga, M. C. (2017). Handbook of clinical psychopharmacology for therapists (8
-
: Gwartney, J. D., Stroup, R. L., Sobel, R. S., & Macpherson, D. A. (2018). Economics: Private⏎and public choice (16
-
photography studio that Dalmau opened in 1854.⏎Download data is not yet available.⏎How to Cite⏎Cuenca, C. (2019). The The “Animated Polystereoscope
-
provides an example of a revised “show-me” sentence and six exercises for students to practice.|⏎Baines, L., & Kunkel, A. (2010). Going bohemian
-
.⏎McLaren, P., & Farahmandpur, R. (2001). Class, Cultism, and Multiculturalism. Mult
-
., Lo, C.-M., Chan, K. S.-C., Chan, D. W.-O., & Chung, K. K.-H. (2014, July). Effectiveness of a tiered intervention model in
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: the numbers. I know that the numbers are the numbers, the numbers. So, if you have a number of coins, then, for example, they are in a certain quantity, and I know that I have them. So, the
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
date. The kids love to play with these sound shakers. So, now it’s your turn to make this easy
-
time should focus on four critical areas: (1) developing understanding of and applying proportional relationships; (2) developing understanding of operations with rational numbers and working with expressions and linear equations; (3) solving problems involving
-
were laser cut (bottom right).⏎Task: For this final project exercise, students were given the constraint of working with a common and uncommon design material. The team highlighted in this example were given wood
-
the formulas that we have used so far.⏎- Just be careful about using the correct sign (positive or negative) depending on the problem.⏎Examples of Calculations with Gravity⏎Example 1: A ball is thrown up into the air at an
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
- L5:2952 excites
- L5:1929 weakly suppresses
- L6:903 weakly suppresses
- L5:2081 weakly excites
- L3:1708 weakly suppresses
- L6:2409 weakly excites
- L5:1810 weakly suppresses
- L5:2934 weakly excites
- L3:680 weakly suppresses
- L6:2042 weakly suppresses
- L6:2827 weakly excites
- L1:608 weakly excites
- L1:1888 weakly suppresses
- L3:2480 weakly excites
- L5:1368 weakly suppresses
- L4:1829 weakly excites
- L1:2467 weakly suppresses
- L6:2434 weakly suppresses
- L0:223 weakly suppresses
- L4:3049 weakly suppresses
- L2:2892 weakly suppresses
- L4:2533 weakly excites
- L1:2801 weakly excites
- L2:892 weakly suppresses
Feeds later layers
- L8:685 excites
- L21:1907 suppresses
- L22:2312 suppresses
- L12:2888 weakly suppresses
- L12:2737 weakly suppresses
- L9:1909 weakly suppresses
- L24:1984 weakly suppresses
- L8:2960 weakly suppresses
- L23:461 weakly suppresses
- L24:2941 weakly excites
- L11:99 weakly suppresses
- L10:552 weakly excites
- L8:2357 weakly suppresses
- L9:521 weakly excites
- L10:2624 weakly suppresses
- L10:2542 weakly excites "is" after "it" in advice phrasing
- L13:1729 weakly suppresses
- L17:2539 weakly suppresses
- L9:1964 weakly suppresses
- L12:1107 weakly excites
- L16:1553 weakly suppresses
- L20:1955 weakly suppresses
- L22:2139 weakly suppresses
- L23:1837 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
- сен
- 正品
- Comparer
- psychiatrist
- 早早
- 购车
- IFICATION
- Comparison
- 正版
- 独立董事
Suppresses
- aths
- manipulating
- mk
- gon
- alie
- (Scene
- lửa
- manip
- omon
- 还想