the "10" in DOI prefixes
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. 21 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 the "0" in the "10." that starts a digital object identifier, the code in academic references like "doi:10.1016/...". It tends to be quiet on other numbers. Turning it up gave unrelated text, so the steering did not reveal more about it.
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.
-
<|endoftext|>- freely available⏎Sustainability 2013, 5(7), 3006-3008; doi:10.3390/su5073006⏎I
-
cript No. AAAJMR-22-82571(R); Published: 29-Nov-2022, DOI:10.35841/ aaajmr-6.11.
-
head injury⏎Ann Afr Surg. 2018;15(2):57-61 DOI:http://dx.doi.org/10.4314/aas.v15i2.5⏎©
-
calculated that some parts of the moon spend about a quarter of the moon's orbit protected by Earth's magnetosphere (Geophysical Research Letters, DOI: 10.1029/2007GL03050
-
ags in the restoration of grassland butterfly communities: a multi-site assessment. Biological Conservation, 155. 50-58. 10.1016/j.biocon.2012.05
-
iodiversity and Conservation (2004) 13: 2511. doi:10.1023/B:BIOC.0000048
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: If you are a human, then the number of ways to select a single valid way to choose a set of possible ways to choose a set of possible ways to select a set of possible ways to select a set of possible ways to select a
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
As were extracted from specimens using TRIzol reagent. Qiagen's miScript System was used to synthesize cDNA and perform qPCR (Figure 1). Endogenous miRNAs are not polyadenylated, therefore during the
-
, has been the focus of intense attention in plant science research for its wide-reaching implications to growth and development (Figure 1). At the heart of this clock is a feedback loop of gene expression known as
-
PCR product amplification is detected by the level of fluorescence emitted by SYBR Green, a dye which intercalates into double stranded DNA. Specific miRNA prim
-
chemical of paraffin-embedded human ovary tumor using 10201-2-AP (ubiquitin antibody) at dilution of 1:50 (under 40x lens).⏎Protein ubiquit
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
- L9:1662 excites
- L20:96 excites
- L3:992 excites
- L19:1426 excites
- L15:2767 weakly excites
- L4:2867 weakly excites
- L20:150 weakly excites
- L10:378 weakly suppresses
- L3:1259 weakly suppresses
- L3:1091 weakly excites
- L7:399 weakly excites laboratory and research facility terms
- L2:94 weakly suppresses
- L9:1805 weakly suppresses
- L15:1841 weakly excites
- L8:2952 weakly excites
- L3:37 weakly suppresses
- L6:2904 weakly suppresses
- L18:2054 weakly suppresses
- L5:2834 weakly suppresses
- L4:40 weakly suppresses
- L8:1834 weakly excites
- L0:1263 weakly suppresses
- L20:1136 weakly suppresses
- L8:2213 weakly excites
Feeds later layers
- L22:902 weakly excites
- L22:1474 weakly suppresses
- L25:671 weakly excites
- L23:2061 weakly suppresses
- L22:1948 weakly excites
- L23:1484 weakly excites
- L22:3052 weakly excites
- L23:2755 weakly excites
- L25:1244 weakly excites
- L25:2739 weakly suppresses
- L24:1769 weakly suppresses
- L24:526 weakly suppresses
- L25:2820 weakly suppresses
- L22:117 weakly suppresses
- L22:2757 weakly excites
- L25:2941 weakly excites
- L26:2143 weakly excites
- L22:2108 weakly excites
- L26:966 weakly excites
- L22:2140 weakly excites
- L24:867 weakly excites
- L26:2351 weakly excites
- L24:2228 weakly excites
- L25:1673 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
- ób
- .forName
- ода
- á
- histó
- ('${
- <![
- Про
- 作息
- vocals
Suppresses
- ucks
- uzu
- asel
- Ksi
- setDisplay
- ardless
- thaw
- osci
- inned
- ềm