'organisms' and 'Code' in reference contexts
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. 6 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 responds to a handful of words that name things in science and law, such as 'organisms', 'Code' in 'United States Code', and 'object'. Turning it up did not produce a clear topic in the writing.
Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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.
-
§ 1395 tells us that this cited federal statute can be found in Title 42, Section 1395 of the United States Code.⏎Code: In general, the term "code" refers to the main body
-
to the same ___________.⏎In the scientific name of the white oak, Quercus alba, Quercus is the __________________.⏎Two groups of organisms that are farther from each other on a cladogram share __________ derived traits than
-
Revised Statutes). The statutes that are published in a state’s code are grouped by subject matter into Titles, as in Title 11 of the United States Code (bankruptcy laws). In some states, including California, Texas and New York
-
buses.⏎- Cars, trucks and motorcycles produce three times the pollution of buses.⏎These comparisons assume the vehicles are not filled to capacity on average. Trains come out even better in comparison to planes when both are full. (Note that Amtrak
-
history.⏎How did Aristole classify life? What was the problem with his system?⏎Scientist should use scientific names because common names can be misleading due to organims having multiple names. One type of species can live all over the world and can
-
An element can be represented on several diagrams. You can quickly locate which diagrams it occurs in throughout the model,
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: a part of a part of the system. I (I) can't't. And A.2553355 256, 700, 700, 700
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
- 4 Clinical Trials⏎- 5 Cohort (Longitudinal) Studies⏎- 6 Case-Control Studies⏎- 7 Mesures of Morbidity and Mortality⏎Fundamentals in Epidemiology Course Description⏎Fundamentals of
-
%, respectively, of all cases reported by EU/EEA countries. 31 deaths were reported for this period.⏎Of 13,783 cases with known vaccination status, 87% were unvaccinated,
-
1971 and 9.8 million in 2001, according to a report published in the March 11 issue of Morbidity and Mortality Weekly Report.⏎And the number of cancer survivors continues to grow because
-
journey in a computer graphics world (C.A. Pickover). Part II. Attractors. Automatic generation of strange attractors (J.C. Sprott). Attractors with dueling symmetry (C.A. Reiter). 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
- L5:216 suppresses
- L4:200 excites
- L5:264 suppresses
- L5:1939 suppresses
- L5:2709 suppresses
- L4:2695 weakly excites
- L4:867 weakly excites
- L4:190 weakly suppresses
- L5:2190 weakly suppresses
- L5:314 weakly suppresses
- L3:393 weakly excites
- L5:2937 weakly suppresses
- L4:215 weakly excites
- L4:151 weakly suppresses
- L4:517 weakly excites
- L4:2103 weakly suppresses
- L3:390 weakly excites
- L5:28 weakly excites
- L4:272 weakly excites
- L1:2699 weakly suppresses
- L5:60 weakly excites
- L4:2459 weakly suppresses
- L4:37 weakly suppresses
- L2:480 weakly suppresses
Feeds later layers
- L10:1073 suppresses
- L15:2846 suppresses
- L8:120 excites
- L8:1311 excites
- L7:769 suppresses
- L15:1044 weakly suppresses
- L14:1760 weakly suppresses
- L10:1507 weakly suppresses
- L11:2402 weakly suppresses
- L9:620 weakly suppresses
- L8:2240 weakly suppresses
- L12:2098 weakly suppresses
- L9:707 weakly suppresses
- L13:705 weakly excites
- L7:116 weakly suppresses
- L9:1289 weakly suppresses
- L7:62 weakly suppresses
- L17:2111 weakly suppresses
- L14:1587 weakly suppresses
- L8:845 weakly suppresses 'too' meaning 'also' at clause end
- L11:2685 weakly suppresses
- L16:3042 weakly suppresses
- L14:2846 weakly suppresses
- L12:608 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
- cbd
- cb
- cbc
- renders
- �
- sh
- sel
- mounts
- md
- .RequestBody
Suppresses
- phia
- async
- Exited
- Caught
- extrême
- Moder
- 都要
- 厚厚的
- esters
- gone