opening parenthesis before phone or code
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. 10 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 an opening bracket that comes before a phone area code or a short aside, like the '(' in '(318) 3' or '+81-(0)48'. It goes strongly negative on brackets that open a year or a place name in a citation. Turning it up did not make the model's writing about phone numbers.
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.
-
This answer is for general informational purposes only and is not a substitute for professional medical advice.⏎If you think you may have a medical emergency, call your doctor or (in the United States) 911 immediately. Always seek the advice of
-
-(0)48-503-9576 / Fax: +81-(0)48-503-9609⏎Ms.
-
benefit from an angioplasty, schedule an appointment with Dr. Gonzalez by calling the Cardiovascular Diagnostic Center, APMC, in West Monroe, LA, at (318) 3<|endoftext|>The Domain Name
-
Use the alphabetical glossary listing to find a short definition to a medical or dental term. If you would like to know more, please remember to call us on (02) 9420 5050. Our friendly
-
(core) city ; Greek ἄστυ asty ) with the associated surrounding area ( χώρα chōra , [the] ' Chora '). Its residents were not legally distinguished from the residents of
-
ore, find out the cause with the help of an overnight sleep study in New Jersey at Sleep Dynamics. Schedule a consultation with a sleep medicine specialist today by calling (848) 217-
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 can say, as a note of instruction, as a note of encouragement). Note the phrase in 1st paragraph: maybe the user is to include the first paragraph: PP) the first paragraph: P) maybe the user is to
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
ter, Taslima and Simonovic, Slobodan P., "A General Overview of Multi-objective Multiple-participant Decision Making for Flood Management" (2002). Water Resources
-
the United States against Spain. In Article III, the treaty stated that: "The boundary line between the two countries. . . shall begin on the Gulph [sic] of Mexico, at the mouth of the river Sabine, in the
-
an active training ground for American troops.⏎“Florida’s mild climate and flat terrain allowed for year round training for aviation,” says Graetz. “Camp Blanding [near Starke] is now a National Guard Camp. During World War II,
-
the air space with a soft polymer. The result was a continuous reinforced carbon nanotube composite.⏎“People have been working for more than ten years intensively [on carbon nanotube composites],” Suhr said. “Nobody had
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
- L9:339 excites
- L8:354 excites opening parenthesis before a citation year
- L9:100 excites
- L3:57 excites
- L0:1252 suppresses
- L6:144 weakly excites
- L4:2309 weakly excites
- L6:7 weakly suppresses
- L6:2958 weakly excites the colon after Note at the start of a note
- L4:2742 weakly excites
- L0:992 weakly excites
- L8:1081 weakly suppresses
- L8:2269 weakly excites
- L4:2111 weakly suppresses
- L9:2388 weakly excites
- L7:1049 weakly suppresses
- L4:701 weakly suppresses
- L0:847 weakly suppresses
- L8:2608 weakly excites
- L2:814 weakly suppresses
- L8:2075 weakly suppresses
- L0:1719 weakly suppresses
- L7:127 weakly suppresses
- L2:2125 weakly suppresses
Feeds later layers
- L16:264 suppresses opening parenthesis in citations and asides
- L13:1871 suppresses
- L17:304 suppresses opening parenthesis after a claim
- L17:1509 suppresses
- L16:1513 suppresses opening parenthesis before a numbered list
- L24:289 excites
- L19:453 suppresses
- L12:2803 weakly suppresses
- L18:682 weakly suppresses
- L19:2251 weakly suppresses opening parenthesis after a name
- L14:2228 weakly excites
- L14:2142 weakly suppresses
- L12:2745 weakly suppresses
- L15:2747 weakly suppresses
- L22:144 weakly suppresses
- L21:1807 weakly suppresses
- L17:1976 weakly suppresses
- L11:226 weakly excites
- L18:1927 weakly suppresses
- L17:2063 weakly suppresses
- L16:169 weakly suppresses
- L14:2402 weakly suppresses
- L12:1897 weakly excites
- L14:1727 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
- xious
- 大城市
- olation
- IMIZE
- esty
- 经济损失
- 级以上
- _global
- 百万
- 级
Suppresses
- cie
- ữa
- kd
- quam
- Reminder
- 称为
- )"); ⏎
- Sus
- obar
- jr