blank spot where a clock time comes next
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. 23 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 empty space right after a date or a 'at' phrase, just where a clock time such as 4:39 PM or 2:00 p.m. is expected next. Turning it up did not produce a clear theme in what the model wrote.
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.
-
Previous Article:||World match.|⏎|Next Article:||Ukraine's Orange Revolution.|<|endoftext|>posted on November 17, 2014 23:23⏎Scientists from the WCS (Wildlife Conservation Society),
-
.com⏎Fri Sep 10 00:10:22 CEST 2010⏎On 9/9/10 4:39 PM, Baba wrote:⏎> In below code "
-
AdminStaff1 · Views: 1,130⏎Contributed by: AdminStaff1 · April 10, 2008 @ 12
-
pediatric and 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 393-39
-
photos and video footage from NASA satellites to discuss the visual phenomenon known as the Northern Lights.⏎- Sat., Mar. 2, 2013, 10:30 AM⏎- Terrace Gallery⏎Please use the event calendar
-
.⏎Though, I guess, it's not the first time it's happened...<|endoftext|>Last updated on September 17th, 2022 at 02:16 pm⏎Synodontis catfish are one of 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 more local, the more local, the more local, the more local... new local and new local and new local and new local and new local... and I am thinking of a new local to add. I need to create a
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
148 ft (350 m)|⏎|Time zone||Eastern (EST) (UTC-5)|⏎|• Summer (DST)||EDT (UTC-4)|⏎|GNIS feature ID||097
-
'tray' and covering it with paper.⏎My cousin was 24 years old on Friday April 5th in 1974. On what day⏎of the week was she born?⏎A tetromino is made up
-
11)|| 0.805⏎very high · 43rd⏎|Currency||Euro (€) (EUR)|⏎|Time zone||EET (UTC+2)|⏎|-||Summer (DST)||E
-
Windows 11 users will benefit hugely from learning how to zip their bulky files and if the need arises to unzip them.⏎Also read: How To Change Date And Time In Windows 11⏎How To Zip Your Files In Windows 1
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
- L20:1335 suppresses
- L22:2068 suppresses
- L19:2242 suppresses
- L9:811 suppresses
- L21:1185 excites
- L20:1355 suppresses
- L20:2792 excites
- L12:2338 excites
- L21:375 suppresses
- L22:2982 excites
- L22:2445 excites
- L5:55 suppresses
- L4:2826 excites
- L16:1668 suppresses
- L21:2290 excites
- L14:1088 excites
- L4:2570 suppresses
- L22:1536 weakly suppresses
- L17:1872 weakly excites
- L4:586 weakly suppresses
- L9:1057 weakly excites
- L20:1155 weakly excites
- L18:1511 weakly suppresses
- L4:2497 weakly suppresses
Feeds later layers
- L27:1288 suppresses
- L24:637 suppresses
- L25:1832 suppresses the word 'local'
- L24:1904 weakly suppresses
- L24:950 weakly suppresses
- L24:2654 weakly excites
- L27:679 weakly suppresses
- L26:2174 weakly suppresses
- L27:1207 weakly suppresses
- L27:710 weakly suppresses
- L25:1021 weakly suppresses
- L24:1023 weakly excites
- L24:1069 weakly excites
- L27:391 weakly excites
- L26:2435 weakly suppresses
- L25:2922 weakly suppresses
- L24:3022 weakly excites
- L26:2610 weakly excites
- L24:188 weakly excites
- L25:1872 weakly excites
- L24:1463 weakly excites
- L25:224 weakly excites
- L25:155 weakly suppresses
- L26:1274 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
- liğini
- iameter
- ApplicationBuilder
- /embed
- progressbar
- راء
- xAB
- xAA
- 台上
- iores
Suppresses
- Local
- local
- local
- Local
- 本地
- _local
- →Local
- →local
- -local
- LOCAL