cutting and dividing in instructions
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 lights up on words about cutting or splitting things apart, like 'cut out', 'break up' or 'divided between', mostly in craft and how-to text. Its opposite side fires on phrases about breaking a job into small steps. Turning it down made the model drift into step-by-step planning, which fits that opposite side.
Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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.
-
level, such as transplanting seedlings.⏎Hoes and Mattocks⏎Hoes are tools used to create furrows in the garden for planting; to break up hard, clumpy soil; and to remove weeds. Hoes come in
-
glue the printed pages to cardboard (like old cereal boxes) and then cut out the cards and boards.⏎The Rules: To play, you need to print out the deck
-
to trace the person's shadow onto the⏎poster board. After the image is traced, turn the lights on and cut around⏎the tracing. Using the cut-out as a pattern, trace the person's shape onto⏎a dark piece of paper
-
and Craft Projects for Presidents Day⏎Copy silhouettes of Washington and Lincoln onto sheets of cardstock. Give⏎each child a sheet. Let children cut the sheet into 4-12 pieces (depending⏎on ability level), then re
-
the students. We recommend reading this ceremony using a narrator and two student readers; their parts are marked below. The texts titled "Student" are to be divided between the two readers. Solemn background music is appropriate. There are two notes in
-
um first – not just skinning the poor animal, removing the hair, stretching, drying and processing the skin into parchment, but also preparing the sheets, pricking
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 thought." So, for example, the first step is to create something that can be done in smaller pieces to get the most out of the best. Then, the first step is to create a piece of work that is
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
time to market and minimizes post-release bugs. Instead of a complex implementation of global tasks, the team following the principles of the Agile development cycle performs tasks in small sprints lasting 1-4 weeks, depending on the specifics of the project
-
student during in-class assignments.⏎- Use more than one way to demonstrate or explain information.⏎- Break information into small steps when teaching many new tasks in one lesson (state objectives, review previous lesson
-
landscape may seem daunting, but by breaking the process down into smaller steps, organizations can leverage their security program to achieve business objectives. One critical component
-
know how to handle unlike denominators or how to regroup or to borrow with mixed numbers? Take a deep breath and let us take it one step at a time. Click and begin.⏎GCF and LCM by Prime Factorization⏎A
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
- L6:2746 weakly excites
- L6:1724 weakly excites
- L3:1413 weakly excites
- L4:1598 weakly excites
- L4:2131 weakly suppresses
- L6:1465 weakly excites
- L6:386 weakly suppresses
- L6:2793 weakly excites
- L2:657 weakly excites
- L6:1799 weakly suppresses
- L2:2974 weakly excites
- L2:1368 weakly excites course and business document nouns
- L6:833 weakly suppresses
- L5:1968 weakly suppresses
- L0:2402 weakly excites
- L5:3013 weakly suppresses
- L5:448 weakly excites
- L4:1545 weakly suppresses
- L6:2451 weakly suppresses
- L3:2159 weakly suppresses
- L2:1606 weakly excites
- L2:3029 weakly excites
- L6:1390 weakly excites
- L6:1481 weakly excites
Feeds later layers
- L12:2621 suppresses
- L22:1856 suppresses
- L19:1924 suppresses
- L9:1106 suppresses
- L8:2688 suppresses
- L14:2426 suppresses
- L21:828 suppresses
- L19:2500 suppresses
- L21:2067 suppresses
- L9:2505 suppresses
- L10:2059 suppresses
- L12:2840 suppresses
- L10:1559 suppresses
- L23:292 weakly excites
- L9:1739 weakly excites
- L11:1993 weakly suppresses
- L20:2112 weakly suppresses
- L8:2798 weakly suppresses
- L23:1621 weakly suppresses
- L17:1168 weakly suppresses
- L18:1168 weakly excites
- L9:2292 weakly suppresses
- L13:2374 weakly suppresses
- L11:2976 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
- argar
- omen
- aptor
- вести
- FieldType
- 公网
- ären
- ialis
- ivar
- !.⏎⏎
Suppresses
- 分割
- divisions
- split
- partitions
- ProgressDialog
- цы
- 磊
- .Concurrent
- fragments
- 分成