'of' after a process noun
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
This neuron fires on 'of' when it follows a noun that names an action or change, like 'citation of sources', 'warming of the climate' or 'changing of the seasons'. Turning it up gave text with no clear theme.
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.
-
duction defects (abnormalities of the electrical activity and the coordination of proper contractions of the heart); (0)cular hypertelorism (widely-spaced eyes); (P)ulmonary stenosis (
-
Fletcher’s lesser-known plays, The Woman’s Prize, evidences. (I must get round to reading this one, it’s a sequel to The Taming of the Shrew
-
iographic conduction defects; (O)cular hypertelorism; (P)ulmonary stenosis; (A)bnormalities of
-
of six reliable references, two of which should be peer-reviewed articles.⏎Write a 7–8-page paper in Word format. Apply APA standards to citation of sources. Use the following file naming convention: LastnameFirstInitial_M5_A
-
to Schwaller de Lubicz, this mysterious dismantling of temples, found at Karnak and numerous other places in Egypt, has to do with the changing of the astrological cycles. The sup<|endoftext|>You are here: PureEnergy
-
goals. In addition, the completion of the process will pave the ground for another messier civil war and defeat of the convoy of 19-year democracy in Afghanistan.⏎Irresponsible and early
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 bought this year • this year • this year • this year • this year • this year • this year • this year • this year • this year • this year • this year
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
some minimal tools, you just start.⏎So let’s have a look at what you need and how to get started.⏎Seed Starting Supplies⏎To get your seed starting done all at
-
presses without any consideration given to centering the load.⏎Progressive and transfer dies produce many types of stamped parts. Consolidation of multiple parts into a single stamping and down-gauging material thickness to reduce component weight increases part complexity. This
-
age group.⏎DOSAGE AND ADMINISTRATION⏎Preparation For Administration⏎Shake the syringe vigorously before administering the vaccine and shake the multidose vial preparation each time before withdrawing a dose of vaccine.⏎Inspect FLUVIRIN® sy
-
matter that shoud be added to the compost heap – washing up water!⏎Many of us want to be more environmentally friendly but each day we happily pour grease laden washing up water down the drain. Greasy water causes major problems in the
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
- L4:261 suppresses
- L6:1684 suppresses
- L6:322 suppresses
- L5:384 suppresses
- L4:253 suppresses
- L6:31 excites
- L3:753 excites
- L2:1006 excites "of" after a process noun
- L4:428 suppresses
- L5:2218 suppresses
- L5:184 excites
- L6:141 suppresses
- L3:349 suppresses
- L2:172 excites
- L5:1799 suppresses
- L4:2511 suppresses
- L5:2509 excites
- L3:110 excites
- L5:129 weakly excites
- L1:911 weakly excites
- L6:690 weakly excites
- L1:192 weakly suppresses
- L6:105 weakly excites
- L2:1354 weakly excites
Feeds later layers
- L12:2287 excites
- L15:835 suppresses
- L9:388 excites
- L14:2909 excites
- L11:1309 suppresses
- L16:2023 excites
- L13:91 excites
- L9:1240 excites
- L9:326 excites
- L9:1903 excites
- L10:1462 excites
- L8:199 suppresses
- L14:1058 excites
- L15:2685 excites
- L17:1177 excites
- L8:89 suppresses
- L9:152 suppresses
- L9:3053 suppresses
- L10:292 excites
- L12:1342 weakly excites
- L19:2893 weakly excites
- L10:998 weakly excites
- L21:1932 weakly suppresses
- L20:465 weakly 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
- responsiveness
- missive
- se
- Resolved
- adherence
- responsive
- respective
- eventual
- __("
- loo
Suppresses
- ABOUT
- kers
- 年来
- Fridays
- /she
- 给她
- emed
- 舍不得
- /IP
- 投行