mixed, web boilerplate phrases
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 fairly consistent: several of them 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 scattered words in web page clutter, such as 'ago' in 'a year ago', 'by' in 'Posted by', and 'on' in 'first appeared on'. Its strongest negative cases sit right at document breaks and login prompts. Turning it up did not give a clear, consistent theme.
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 balloons.⏎Submit a review<|endoftext|>1 year ago⏎Verified Sherpa Tutor⏎If a liquid has a definite volume, what
-
the all-loving creator who meets every human need. In Psalms we read, “Give thanks to the Lord, for he is good; his love endures forever. Let the redeemed of the Lord tell their story. . . . Let
-
ane) and is a natural by-product of gas<|endoftext|>DataTableCollection.Item Property (String)⏎Gets the DataTable object with the specified name.⏎Assembly: System.Data (in System.Data.dll)⏎If a DataTable name exists that matches the case of the search
-
world. Proverbs 31:8-9 -- a noble man stands up for what is right and shows concern for others. Such are the sayings of King Lemuel, a noble man belonging to God. Many biblical scholars believe that
-
Holman Christian Standard BibleJesus told him, “Foxes have dens, and birds of the sky have nests, but the Son of Man
-
Retrieved from https://youtu.be/qlSxPouPCIM⏎The post 1929 Stock Market Crash and the Great Depression. first appeared on COMPLIANT PAPERS.⏎The post 1929 Stock
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: The main thing that you have to know, if it's a given. So, you have to be a given, and then be a person, and then be a whole, and then you can't do it without a hat, a
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
,0<|endoftext|>Guidelines for Restoring Cultural Landscapes⏎Identifying, retaining and preserving
-
, 2012<|endoftext|>Some weather conditions are simply not fun to ride in. As cyclists, windy days
-
weaving and entrance ramp studies produced some very significant results, which are discussed in the report and will contribute to overall knowledge of freeway operation.⏎You must be logged in to post a
-
liu.<|endoftext|>Bond Price Calculator⏎With this bond price calculator, we aim to help you calculate
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
- L9:147 suppresses
- L5:414 suppresses
- L7:1212 suppresses
- L10:169 excites
- L11:483 suppresses
- L9:1301 excites
- L1:2017 suppresses
- L3:604 excites
- L7:105 excites
- L6:1003 excites
- L5:1303 weakly excites
- L8:24 weakly suppresses
- L11:17 weakly suppresses
- L4:2500 weakly excites
- L9:81 weakly suppresses
- L9:253 weakly excites
- L4:625 weakly suppresses
- L9:1798 weakly suppresses
- L4:335 weakly suppresses
- L3:323 weakly excites
- L6:1576 weakly suppresses
- L10:1545 weakly suppresses
- L8:254 weakly excites
- L4:2289 weakly suppresses
Feeds later layers
- L16:248 excites
- L14:267 suppresses
- L13:283 suppresses
- L15:1304 suppresses
- L14:1121 suppresses
- L18:738 suppresses
- L14:84 suppresses
- L14:2556 suppresses
- L19:446 suppresses
- L19:251 suppresses
- L17:892 suppresses
- L13:2188 suppresses unclear: weak, scattered tokens
- L16:275 suppresses
- L13:2946 suppresses
- L15:1607 suppresses
- L19:1273 suppresses
- L19:802 excites unclear: word before a company or brand name
- L21:105 suppresses
- L16:1654 suppresses
- L17:841 suppresses
- L25:2284 suppresses
- L20:151 suppresses
- L13:450 excites
- L20:573 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
- 发生
- луш
- endency
- ăn
- ц
- ascript
- Annotation
- annotation
- .getContentPane
- icon
Suppresses
- →del
- tür
- eil
- 五一
- functional
- terdam
- ויות
- beit
- reno
- embar