day number after a month name
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. 12 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 day number that comes right after a month name, like the 15 in "June 15" or the 28 in "May 28th". When it was turned up, the written text did not show a clear change in topic or style.
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.
-
, biotech⏎companies activate a patients’ own T cells so that they attack cancer cells.⏎Honoring International Researchers on World Cancer Day⏎Every year on February 4, World Cancer Day unites the world’s population in the fight against cancer
-
and decision-making scenarios to help define the process that people go through when formulating solutions to problems.⏎Herbert Simon was born in Milwaukee, Wisconsin, on June 15th, 1916. He went to the Milwaukee public
-
. The staff housing located not far from the main buildings are not secluded<|endoftext|>Do you know about Menstrual Hygiene Day, which is observed annually on May 28th?⏎This Saturday is Menstrual Hygiene Day, which will
-
appropriate menstrual hygiene management around the world. Menstrual Hygiene Day (MHD, MH Day in short) is a yearly basis awareness day held on May 28 to emphasize the need for excellent menstrual hygiene management (
-
everyone in our country because it is not a normal person history it is legend’s biography that should be read by all.ere we are giving short speech on October 2nd. Gandhiji full name is Mohandas Karam Chand Gandhi he
-
was born in Montpelier, Vermont, on December 26, 1837. A hero of the Civil War and
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: "I" (myself) in the past, the past. [inserted in the past, but not sure where I was at the moment." (myself) [in the past" ... [end of this passage] The
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Indus Valley Civilisation of Harappa and Mohenjo-daro from the Bronze Age. Nowadays, the Thar
-
statue from Nimrud’s Temple of Ishtar, decorated with cuneiform engravings detailing the life of King Ashurnasirpal II, who reigned from 883 to 8
-
- Who produced the first really useful TV system?⏎Answers of Popular Inventions Quiz Questions⏎- William Caxton⏎- The Thermometer and the Telescope⏎- Paper, Printing, Gunpow
-
⏎The major mountains are the Mt.Everest, Mt.Lhotse, Cho Oyu, Nuptse, Pumori, Ama Dablam
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
- L11:974 excites
- L11:12 weakly suppresses
- L9:181 weakly suppresses
- L11:1123 weakly excites
- L10:2991 weakly excites
- L5:509 weakly excites the word 'even' after a clause
- L9:2680 weakly excites
- L11:2427 weakly excites
- L2:497 weakly excites
- L9:643 weakly suppresses
- L9:171 weakly suppresses
- L8:2647 weakly excites
- L8:824 weakly excites
- L8:1554 weakly excites
- L11:54 weakly suppresses
- L5:21 weakly excites
- L4:2263 weakly excites
- L10:1847 weakly excites
- L0:1829 weakly suppresses
- L7:2362 weakly suppresses
- L11:383 weakly excites
- L7:1587 weakly excites
- L10:2218 weakly suppresses
- L10:2117 weakly suppresses
Feeds later layers
- L14:2025 excites
- L13:1215 excites
- L14:2198 excites
- L13:291 suppresses
- L14:2992 excites
- L13:1380 weakly excites
- L13:1166 weakly suppresses
- L13:1991 weakly excites
- L17:2523 weakly excites
- L16:8 weakly suppresses
- L16:498 weakly excites
- L16:83 weakly suppresses
- L13:888 weakly suppresses
- L14:1038 weakly suppresses
- L23:348 weakly excites
- L14:1983 weakly suppresses
- L16:108 weakly suppresses
- L15:45 weakly suppresses
- L16:3062 weakly excites
- L16:137 weakly suppresses
- L13:1966 weakly excites
- L16:142 weakly suppresses
- L16:44 weakly suppresses
- L14:1131 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
- ures
- erah
- URES
- ImageSharp
- egra
- 标准
- Di
- iram
- 缁
- IVO
Suppresses
- doubt
- stakes
- itet
- ไหว
- ัย
- 糙
- át
- thead
- Textbox
- taking