'of' after a number or amount
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. 0 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 responds to the word 'of' when it follows a number or measurement, like '22 of them' or '1/2 of'. The steered outputs did not show a clear effect tied to this pattern.
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.
-
385 of oats; 103,434 of wheat; 88
-
= 2 x 1/4 or 1/4 is 1/2 of 1/2<|endoftext|>Messier 52 (also known as M
-
1/500 of a second, the PWM output will produce a pulse. The length of this pulse
-
.6 square miles (250 km2) of it is land and 0.6 square miles (1.6 km2
-
requires only around 3.5-5 kWh/m3 of electricity.⏎According to the International Renewable Energy Agency (IRENA), desalination
-
0 with 997 athletes (22 of them women) from 24 countries competing in 95 events. The
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: The question is, how do you make a sound in a way that you can make a sound of a certain kind of music that is a certain kind of a music. So I have to think about this. I'm not sure if I
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
most efficient level of power for the system in place.⏎<|endoftext|>Serious diseases, including diabetes, cancer, and eating disorders, can manifest as signs and symptoms inside of the mouth.⏎Learn what those dental words mean.⏎Get dental news feeds delivered directly
-
. Many students considering out-of-state colleges also take the SAT. The state only receives statewide score reports. For
-
comprehend the techniques; members of instructions want a step-by-step guide to enable them to perform some techniques.⏎Etymology is a well-known way to initiate meanings out-of acronyms, which happen to be abbreviations noticable due to the fact
-
way of looking at things. This should be obvious, because they are seeing *everything* from a different perspective. For example, when we think seeing the inside of an important component would be exciting and informative, they consider it a threat to their
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
None: this is the first layer.
Feeds later layers
- L19:378 excites
- L2:1640 weakly suppresses
- L3:1760 weakly suppresses
- L18:350 weakly suppresses
- L18:2227 weakly suppresses
- L25:2286 weakly suppresses
- L19:620 weakly excites
- L21:1452 weakly suppresses
- L5:1340 weakly suppresses
- L5:2888 weakly suppresses
- L7:1856 weakly excites
- L3:1259 weakly suppresses
- L2:847 weakly suppresses
- L21:11 weakly excites
- L5:2576 weakly suppresses
- L2:576 weakly excites
- L21:2 weakly excites
- L12:2312 weakly excites
- L7:2792 weakly excites
- L23:1863 weakly suppresses
- L7:710 weakly suppresses
- L13:2765 weakly excites
- L24:1655 weakly suppresses
- L4:3069 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
- ��
- .Call
- _Service
- faction
- quence
- iore
- 习近
- .imwrite
- 各样
- 洑
Suppresses
- azes
- enna
- izzes
- dez
- agog
- quez
- 告诉大家
- 告知
- giveaway
- 告诉