unclear: scattered code and URL tokens
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. 6 of Qwen3-0.6B. Its strongest snippets are mixed: they peak on different tokens. 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 is hard to pin down. Its positive responses are weak and fall on scattered pieces of URLs, software names and technical words, with no shared meaning. Its strongest negative responses are on words like 'function' in code and on a comma after a copyright year. Turning it up gave unrelated text.
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.
-
population growth or technical progress. Production per capita, at date t, is given by⏎where k denotes capital per capita and A is a productivity parameter. Capital depreciates at the rate dt, where dt is a convex function
-
of powerful, dynamic applications and services. For more information on this great topic, please consult:⏎- Sun's Java Server Pages home (http://java.sun.com/products/jsp/)⏎- Westy Rockwell's article on "JSP Tag Libraries
-
EDIS (keyword search "marketing") and on the UF/IFAS Tropical Research and Education Center (TREC) Agricultural Economics Extension website (http://agecon.centers.ufl.edu/) (Evans and Ballen 201
-
http://www.tandfonline.com/pricing/journal/VECE20|⏎When requesting a correction, please mention this item's handle: RePEc:taf:jeduce:v:42:y:2011
-
specific section of an organism’s genome, causing the organism to behave differently and to pass this difference to all their offspring. CRISPR-Cas9 gene drives do this cut-copy-paste during the reproductive cycle to make sure
-
in regular classes.⏎If you experience problems downloading a file, check if you have the proper application to view it first. In case of further problems read the IDEAS help page. Note that these files are not on the IDEAS site. Please
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 am using a simple script to implement a function that takes a list of integers and returns their sum. Function: Compute the sum of all integers in the list """ # This script is for a simple task to calculate the sum
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
. The exact solution can be found by quadrature and is⏎To use MATLAB we first construct an M-file that defines the equation.⏎% filename rhs.m⏎function ydot=rhs(t,y)%compute the rhs of eqns.⏎%for
-
© 2012, Columbia University Press. All rights reserved.⏎More on rheology from Fact Monster:⏎
-
-rolling and calculating outcomes is handled for the players by the software.⏎-- MarkMusante - 11 Sep 2002⏎Influence on IF⏎Adventure and
-
when you have several commands that you will use over and over it is more convenient to create a script that can run all the commands with one message from you.⏎% filename simple.m⏎ini=1.; %set the variable ini to 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
- L2:322 excites
- L3:5 weakly excites
- L2:1784 weakly excites
- L5:228 weakly suppresses
- L5:50 weakly excites
- L2:41 weakly suppresses
- L0:178 weakly suppresses
- L4:2992 weakly excites
- L4:1836 weakly excites
- L3:63 weakly excites
- L0:1224 weakly suppresses
- L2:2399 weakly excites
- L1:2347 weakly suppresses
- L1:1234 weakly suppresses
- L4:2254 weakly excites
- L4:1689 weakly excites
- L1:2683 weakly excites
- L0:1301 weakly suppresses
- L4:1615 weakly excites
- L4:1867 weakly suppresses
- L0:1800 weakly suppresses
- L3:148 weakly excites
- L2:1136 weakly excites
- L5:471 weakly suppresses
Feeds later layers
- L7:2319 suppresses
- L9:114 suppresses
- L9:100 suppresses
- L8:2922 suppresses
- L10:1791 weakly suppresses
- L11:2244 weakly suppresses
- L8:1361 weakly suppresses
- L12:2106 weakly suppresses
- L9:642 weakly excites
- L13:2300 weakly excites
- L10:2794 weakly suppresses
- L13:3005 weakly suppresses
- L22:1703 weakly suppresses
- L7:2430 weakly excites
- L7:2491 weakly suppresses
- L7:730 weakly suppresses
- L27:1254 weakly excites
- L8:1187 weakly suppresses
- L21:1973 weakly excites
- L8:932 weakly excites
- L13:1463 weakly excites
- L9:2017 weakly suppresses
- L12:1805 weakly excites
- L8:1729 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
- عد
- rel
- URE
- ности
- 要在
- 照样
- ность
- nesia
- _second
- 行李
Suppresses
- '],⏎
- '], ⏎
- enna
- oras
- oria
- */,⏎
- Thou
- hot
- bó
- -ton