unclear: weak, end of web page
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 gives only weak signals, mostly where a web page or document ends and a new unrelated one begins. It is hard to see a clear meaning, and the model's written text did not change in an obvious, consistent way when it was adjusted.
Written by Claude from this neuron's activations and steering outputs. Confidence: low. 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.
-
of a lighthouse. Think of interesting words to describe the looks and feel of the lighthouse. If you can only think of boring adjectives to describe it<|endoftext|>In a creative writing classroom, when the class discusses and analyzes a work of fiction
-
anniversary of his death⏎The article you want to access is behind a paywall. You can purchase this article or subscribe to access all the low countries articles.<|endoftext|>Repeated infections can have various causes, ranging in severity from 'worrying'
-
examples of creating markers, setting markers, and moving point to markers:⏎;; Make a new marker that initially does not point anywhere: (setq m1 (make-marker)) ⇒ #<marker in no buffer>⏎m1to point between the
-
the Ninevites to preach repentance (Jon. 3: 3-10).⏎Our Lord Jesus Christ, addressing the Scribes and the Pharisees who demanded a sign from Him, said that no sign would be given except
-
Coverage⏎- 1st Photos of Mars by Curiosity Rover (Gallery)⏎Copyright 2012 SPACE.com, a TechMediaNetwork company. All rights reserved. This material may not
-
:⏎- US English dictionary⏎What do you find interesting about this word or phrase?⏎Comments that don't adhere to our Community Guidelines may be moderated or removed.<|endoftext|>Even the hardy tardigrade might not stand up to climate change⏎T
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: what if we can create a new way to design and implement a solution that is based on software and is built to support complex problems, with a common approach, and a way that is easy to implement, to be maintained, and to be
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
:|| Web page: http://www.tandfonline.com/VECE20|⏎|Order Information:||Web: http://www.tandfonline.com/pricing/journal/VECE20|⏎When requesting a correction,
-
23(2), 1976 pp. 125–135. doi:10.1112/S0025579300008718.
-
Krystal McLear, 32, who serves on a committee for the network that runs the U.S. organ transplant system.⏎The United Network for Organ Sharing is debating some new policies to change that. Among the proposals: A checklist for
-
Relation to “The Giver”⏎Memory is something that happened in the past and something someone experienced and is in
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
- L10:2394 suppresses
- L9:2282 excites
- L9:310 excites
- L9:1660 weakly suppresses
- L11:993 weakly excites
- L5:1603 weakly suppresses
- L4:2165 weakly suppresses
- L11:985 weakly suppresses
- L5:883 weakly suppresses
- L3:1475 weakly suppresses
- L2:1181 weakly suppresses
- L3:203 weakly excites
- L10:935 weakly excites
- L11:1044 weakly suppresses
- L9:2870 weakly excites dictionary definition wording
- L9:894 weakly suppresses
- L11:891 weakly suppresses
- L2:2360 weakly suppresses
- L5:2965 weakly excites
- L11:834 weakly excites
- L5:73 weakly suppresses
- L11:2944 weakly suppresses
- L10:36 weakly excites
- L8:2824 weakly suppresses
Feeds later layers
- L16:978 suppresses
- L15:1859 suppresses
- L14:2050 weakly suppresses
- L15:485 weakly suppresses
- L13:1205 weakly excites
- L16:1629 weakly suppresses
- L16:431 weakly excites
- L22:232 weakly excites
- L13:2661 weakly excites
- L27:1319 weakly excites
- L17:1088 weakly excites
- L15:2127 weakly excites
- L24:82 weakly suppresses
- L16:1759 weakly excites
- L14:1735 weakly suppresses
- L13:2063 weakly excites
- L17:1957 weakly excites
- L23:1112 weakly suppresses
- L13:1088 weakly excites
- L15:2519 weakly suppresses
- L16:735 weakly suppresses
- L21:1625 weakly suppresses
- L24:2335 weakly suppresses
- L16:729 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
- 稳定的
- standen
- perator
- Ratings
- tures
- elled
- 过得
- pane
- .itemView
- favor
Suppresses
- reat
- 魔王
- ree
- 創
- �
- iej
- (side
- aby
- reative
- ige