lowercase 'l' starting a word
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. 27 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 lights up on the letter 'l' when it begins a new word or syllable, like 'lily', 'lathes' or 'lice'. It goes quiet or negative on capital letters such as 'L' in 'Lymphoma'. Turning it up made the text drift into repeated, garbled words, so the steering shows little about the neuron itself.
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.
-
66:108-110)⏎lo vinu, lə vinə “wine”⏎lo sale, rə ssalə, l
-
for it. For example, pressure will be needed when creating deep indents.⏎Lathes are divided into three types: the engine lathe, special purpose lathes and turret lathes. Some variants are placed on benches and others
-
9–1892) used both the one- and two-syllable pronunciations:⏎The purple flower droops; the golden bee⏎Is lily-cradled: I alone awake.⏎Both of these poets rhyme flower and
-
unciation: (-tru). [key]⏎1. Also,luster; esp. Brit.,lustre.a period of five years.⏎2. Rom. Hist.a lustration or ceremonial purification of the people, performed every five years, after the
-
professionals like lawyers and teachers. This undisciplined lot went on to brave bullets, were jailed, lathi-charged and suffered untold humiliations at the hands of the colonial rulers
-
advanced against Egypt by land and by sea. Ramses’ land army checked the enemy’s advance in southernmost Palestine, and the hostile ships were trapped after being lured into the waterways of the delta. Egypt averted conquest by the nor
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: Maybe you change the way you look you form more reactions you form reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions more reactions
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
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 straight forward way to find the greatest common factor
-
igations⏎Published: Sunday, September 28, 1919⏎Leave a Reply<|endoftext|>Combination Therapy May Help Patients With Follicular Lymphoma⏎A new study in The Journal of Experimental Medicine reveals
-
ideal treatment stage for Drinking Water, Wastewater and Water Reuse facilities.⏎UV light does not affect the taste,
-
environmental factors. Many Councils use an estimated useful life of somewhere between 8 and 25 years.⏎- Sealed Road⏎- Surface Treatment⏎- Wearing Surface⏎- Asphalt Overlay⏎- Sprayed Seal Useful Life Discussion⏎-
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
- L23:1760 suppresses
- L22:2646 excites
- L20:417 excites
- L4:1978 excites
- L19:593 weakly excites
- L18:198 weakly suppresses
- L24:2164 weakly excites
- L20:151 weakly excites
- L17:134 weakly suppresses
- L21:252 weakly excites
- L18:365 weakly excites
- L16:95 weakly suppresses
- L24:105 weakly suppresses
- L17:1982 weakly excites
- L3:2052 weakly excites
- L9:476 weakly suppresses
- L26:495 weakly suppresses
- L22:766 weakly suppresses
- L20:52 weakly suppresses
- L0:94 weakly excites
- L22:1383 weakly excites
- L12:1815 weakly excites
- L19:2548 weakly excites
- L0:662 weakly suppresses
Feeds later layers
None: this is the last layer.
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
- nice
- protected
- section
- uke
- ert
- thag
- etect
- curring
- works
- -widgets
Suppresses
- 意外
- дачи
- 日内
- Shib
- 糟糕
- Украины
- 不起
- 淋
- lez
- shy