"et" in "et al." citations
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. 14 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 "et" in citations like "Smith et al.", the short form for "and others" after an author's name. When it was turned up, the model's output drifted into unrelated rambling about questions and problems, so the steering doesn't show a clear effect.
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.
-
fever. The effect on mastoidectomy is unclear, but a clinically important increase seems unlikely. (Level of Evidence: 2c)⏎Sharland M, et al. Antibiotic prescribing in general practice and hospital admissions for peritonsill
-
into human 293T cells. The prominent spikes of the Env protein are visible under hight magnification. (Dewannieux, M. et al., Genome Res. (2006) 16: gr
-
froid and Delzenne (1998) suggest that inulin-type fructans may be⏎helpful for NIDD. Yamachita et al. (1984) observed a noticeable reduction⏎of cholesterolmia
-
influence of aerosols on the global climate.⏎Ault, A.P., et al. “Raman Microspectroscopy and Vibrational Sum Frequency Generation Spect
-
mathematics in the bilingual classroom, when the language of learning and teaching (LOLT) is not the learners’ mother tongue, is a complex issue (Setati et al 2002).⏎The purpose of this essay is to discuss the
-
) and spruce (c. 70 years old) trees at the opposing sides of the plot (Figure 1a, for details see Grams et al. 2021). Six plots were equipped with roofs and assigned
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 want to test the hypothesis in the sample size (H2: α = α, α = α, α = α, α = α, α, α, α = α, α, α, α = α, α,
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
for myself:⏎that for awhile I was the Heodeninga<|endoftext|>This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.⏎As the soil, however rich it may be, cannot
-
THE JEPSON MANUAL (1993)||⏎previous taxon |⏎Jepson Interchange (more information)⏎©Copyright 1993 by the Regents of the University of California⏎AND IS MAINTAINED FOR ARCHIVAL
-
Copyright © 1997, by Random House, Inc., on Infoplease.<|endoftext|>This was an introductory lesson
-
tionary, Creative Commons Attribution/Share-Alike License⏎- n. An underwater explorer.⏎from WordNet 3.0 Copyright 2006 by Princeton University. All rights reserved.⏎- n. an underwater swimmer equipped with
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
- L13:1157 suppresses
- L10:546 excites
- L12:910 excites
- L12:967 excites
- L11:1327 suppresses
- L10:2446 excites
- L10:2112 excites
- L9:944 excites
- L13:1957 weakly suppresses
- L12:2887 weakly suppresses
- L7:211 weakly suppresses
- L6:2056 weakly excites
- L4:174 weakly suppresses
- L3:1918 weakly suppresses
- L5:1542 weakly excites
- L10:475 weakly excites
- L13:1978 weakly suppresses
- L2:831 weakly suppresses
- L7:2448 weakly suppresses
- L13:1635 weakly excites
- L12:1480 weakly suppresses
- L1:1392 weakly excites
- L10:2855 weakly excites
- L0:2711 weakly suppresses
Feeds later layers
- L15:1859 excites
- L16:940 suppresses
- L16:1322 excites
- L18:2235 excites
- L20:2181 excites
- L18:1306 weakly excites
- L18:56 weakly excites
- L15:636 weakly suppresses
- L16:1962 weakly excites
- L18:360 weakly excites
- L16:452 weakly excites
- L16:2867 weakly suppresses
- L18:953 weakly excites
- L16:320 weakly excites
- L16:996 weakly excites
- L21:617 weakly suppresses
- L16:1225 weakly suppresses
- L18:1486 weakly excites
- L15:432 weakly suppresses
- L17:295 weakly excites
- L15:870 weakly excites
- L19:2165 weakly excites
- L17:2126 weakly excites
- L19:237 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
- 彦
- ians
- 輯
- PTY
- icients
- .JSON
- 个多
- -blog
- тал
- арь
Suppresses
- tuyến
- intColor
- Divider
- füg
- 用工
- ToggleButton
- myModalLabel
- ửa
- 추
- uci