colon in clock times
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. 24 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 the colon in clock times like '12:00' or '10:30 a.m.', which shows up in event listings and schedules. Turning it down made the model repeat the word 'five' over and over, so the neuron seems to have something to do with numbers.
Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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.
-
generally eaten between 6:30am and 7:30am and often consists of rice or soup. Lunch is generally eaten between 12:00noon and 1:00pm. Many people eat out,
-
adult hematologists. This common disease will be explored in depth in an Education Program Session today at 7:30 a.m. and again at 2:00 p.m. in rooms 393-396 of
-
Elizabeth Suratt Nature Center in Mill Spring on Saturday, Feb. 23, at 10:30 a.m.⏎Zone of plenty⏎There’s no single key to Pol
-
and a half, the winds picked up dramatically with gusts reaching nearly 44 mph as the storm rolled to the southwest.⏎By 4:30 p.m., the storm’s outflow boundary—where the the thunder
-
to perform basic statistical functions using both a scientific calculator and Microsoft Excel.⏎Saturday, Aug. 23, 2-4:30 p.m.⏎Basic Statistics with SPSS⏎If your major or intended
-
ensure appropriate group numbers.⏎We will also be offering an activity before this one, from 1:00-2:00 PM, (also at the grade 1
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 five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five five
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
, they can be dangerously habit forming and result in grogginess the following day. Here are some safe techniques to promote good sleep hygiene:⏎- Establish and maintain a consistent sleep schedule.<|endoftext|>A new tool to identify the calls of bat species could
-
. But, in most situations, it’s our phones, TVs, and laptops that are impairing our sleep.⏎In healthy adults, our biological clock follows a 24-hour wake-sleep cycle. As the sun rises and daylight gradually enters
-
once again intervene? Will life<|endoftext|>You can help keep your heart and blood vessels healthy by taking steps toward a healthier lifestyle. These healthy habits include not smoking, eating right, exercising regularly, staying at a healthy weight, and getting the screening tests
-
the other hand, are more short-term commitments that a person makes in order to improve their life or reach a certain goal. And Resolutions can be anything from eating healthier to spending more time with family. Resolutions may not have any specific timeline
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
- L23:1207 suppresses
- L23:569 excites
- L19:1943 suppresses
- L23:3011 weakly suppresses
- L23:637 weakly excites
- L17:2851 weakly suppresses
- L22:2729 weakly excites
- L22:2986 weakly suppresses
- L4:808 weakly excites
- L6:1109 weakly excites the word 'data' in database contexts
- L3:179 weakly suppresses
- L15:59 weakly suppresses
- L10:2286 weakly excites
- L15:155 weakly suppresses
- L10:2425 weakly suppresses
- L19:1729 weakly suppresses
- L20:818 weakly suppresses
- L19:1261 weakly excites
- L20:2771 weakly excites
- L21:2738 weakly suppresses
- L12:1355 weakly excites
- L16:2435 weakly suppresses
- L23:200 weakly excites
- L20:1335 weakly suppresses
Feeds later layers
- L26:760 suppresses
- L27:435 suppresses
- L25:2519 excites
- L25:753 suppresses
- L26:739 weakly excites
- L25:2815 weakly suppresses
- L27:391 weakly excites
- L26:976 weakly suppresses the '1' starting a century
- L26:790 weakly suppresses
- L26:2581 weakly suppresses
- L25:2316 weakly suppresses
- L26:327 weakly excites
- L27:2879 weakly excites
- L27:3043 weakly suppresses
- L25:2806 weakly excites
- L25:1268 weakly suppresses
- L26:3064 weakly excites
- L26:1861 weakly suppresses
- L26:2107 weakly excites
- L27:2496 weakly suppresses
- L27:1138 weakly excites
- L26:2699 weakly suppresses
- L25:1005 weakly excites
- L25:1570 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
- ption
- 重量
- icont
- ont
- <Entry
- 科技大学
- Compet
- Return
- abbr
- 镡
Suppresses
- five
- 五百
- fifty
- .idea
- Five
- Five
- -five
- 5
- Fifty
- 倒霉