the word 'course' in course blurbs
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. 11 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
This neuron reacts to the word "course" when it refers to a class or online course, such as "In this course" or "Before attending this course". Turning it up gave text about learning and training, but the links are loose.
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.
-
016.⏎These media resources form part of an on-demand Coursera MOOC run by Frank Cogliano; Professor of American History,
-
including the date of preparation or last revision. The SDS must also state that no applicable information was found when the preparer does not find relevant information<|endoftext|>About This Course⏎Learn the fundamentals of Spark, the technology that is revolutionizing the analytics and
-
lessons⏎- Available on completion⏎- Course author⏎- Programming Tut⏎- They will have an excellent understanding of the fundamentals in Java programming.⏎- In this course you will get a head-start into your Java studies⏎- In this course you
-
⏎Who should attendExperienced programmers who need to learn the syntax and functionality of the SAS LASR Analytic Server and the IMSTAT procedure⏎Before attending this course, you should have experience writing SAS programs or working with other languages such as R
-
elt, rainfall frequency, and the hydrologic cycle. The streamflow estimation procedures presented are the rational method and the unit hydrograph.⏎Upon successful completion of this course, the student will be able to:⏎CIVL 1060
-
NSM2012. You'll have 30 days to complete the course.⏎For the younger driver, the North Dakota Safety Council is also introducing Defensive Driving
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 don't know what I have." "I don't know what I have." I don't know what I have." "Let me have it, let me have it, let me have it, let me have it." "
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
And by doing group project work and presentations, they could inform other classmates and teachers (i.e., project-based and cooperative learning).
-
arts related fields, but in other academic areas as well.⏎Arts education develops students who are self driven and motivated as they are often the toughest critics of their own
-
, they help children explore the outdoors while developing learning skills.⏎- Getting your Bearings⏎- How Artistic are you?⏎-
-
sharing playing time, and respecting rules. They understand that rules are important for everyone, and that without regulations, games would become unfair.⏎Although competition helps children develop problem solving and other skills, it can be both positive and negative. Negative competition
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
- L9:1741 excites
- L10:863 suppresses
- L9:2657 suppresses
- L6:2984 suppresses
- L8:2772 suppresses
- L10:2796 excites
- L6:2991 suppresses
- L6:1250 excites
- L9:1018 excites
- L3:469 weakly suppresses
- L9:395 weakly suppresses
- L7:2603 weakly excites
- L10:1834 weakly excites
- L10:893 weakly excites
- L8:2155 weakly excites
- L10:2519 weakly suppresses
- L9:1681 weakly suppresses
- L10:1690 weakly excites
- L5:354 weakly excites
- L10:2352 weakly excites
- L9:2977 weakly excites
- L3:1155 weakly excites
- L6:1686 weakly excites
- L6:2814 weakly excites
Feeds later layers
- L20:918 excites
- L25:1540 excites
- L18:2533 excites
- L17:792 excites
- L18:2249 excites
- L21:1958 excites
- L20:2884 excites
- L21:1666 excites
- L21:2422 excites
- L21:3035 excites
- L26:387 excites
- L21:1608 excites
- L21:2739 excites
- L16:1857 excites
- L19:1035 excites
- L18:3054 excites
- L26:382 excites
- L18:2104 excites
- L25:641 excites
- L19:1897 excites
- L19:2246 excites
- L26:909 excites
- L21:2991 excites
- L19:2399 excites
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
- training
- training
- .training
- tutorials
- (training
- 教育培训
- 学员
- tutorial
- -training
- 教程
Suppresses
- anch
- urd
- ira
- oper
- 甾
- obody
- arta
- 数据
- odash
- urn