"prices" after economic words
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. 10 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 word "prices," especially in economic news like "world prices," "wheat prices" or "commodity prices." Turning it up made the model write about prices, food, oil and supply, so the idea comes through clearly. The opposite side is weak and hard to read.
Written by Claude from this neuron's activations and steering outputs. Confidence: high. 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.
-
prices for these resources. Despite Peru's strong macroeconomic performance, dependence on minerals and metals exports and imported foodstuffs makes the economy vulnerable to fluctuations in world prices.⏎Peru's rapid expansion coupled with cash transfers and other programs have helped to
-
5.6% per year from 2009-13 with a stable exchange rate and low inflation. This growth was due partly to high international prices for Peru's metals and minerals exports, which account for 55% of
-
Pakistan comes as a drought and wildfires in Russia have cut the harvest there and have helped to drive up worldwide wheat prices.⏎The floods in Pakistan also have destroyed the logistics and transportation systems that make it
-
, stream bank erosion along the Yukon River, as well as uneven management in the Yukon Flats National Wildlife Refuge. People are worried about ever-rising prices for fuels and store-bought staples, and fewer and fewer sources of wage income
-
in Peru’s lucrative mining sector, a move that was opposed by some environmental groups. However, in 2015, mining investment fell as global commodity prices remained low and social conflicts plagued the sector.⏎Peru's free trade policy continued
-
and volatile food prices. These factors compromise responses to climate changes – both ‘normal’ variation and shocks
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 not sure that I would like to see a lot of things." So the first step is to place the object on a surface, and the first step is to place a single object on a surface, so the next step is
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
to something more complicated like a house that is bought, or a lawyer that is hired. You need to know how to count money to guard yourself against being shortchanged. Being able to properly count money is also important due to it being the basis
-
welfare state is socialistic in nature. It is based on the principles of equality and is keen to provide equal opportunity to all. It also aims to ensure equitable distribution of wealth.⏎3.It exercises control over all the economic activities. In a welfare
-
pecoupé, toupée⏎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|>Why travel to Europe to see old ruins when you have some fantastic ones
-
of people gathered for the event. Corner lots sold as high as $1,250.00, and “inside” lots were sold as high as $1000.00. Within four months, these men
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
- L9:2912 suppresses
- L9:2255 suppresses
- L8:2346 suppresses
- L6:2256 excites
- L7:1774 weakly excites
- L8:2478 weakly suppresses
- L5:1202 weakly excites
- L7:2346 weakly excites
- L4:1077 weakly suppresses
- L3:1346 weakly suppresses
- L5:1562 weakly suppresses
- L9:2538 weakly suppresses
- L4:898 weakly suppresses
- L7:2239 weakly suppresses
- L9:569 weakly suppresses
- L5:1129 weakly excites
- L9:1460 weakly excites
- L2:1655 weakly suppresses
- L6:1564 weakly excites
- L1:2397 weakly suppresses
- L4:1816 weakly suppresses
- L7:1612 weakly suppresses
- L2:2800 weakly excites
- L9:2836 weakly suppresses
Feeds later layers
- L11:423 excites
- L19:2725 excites
- L11:587 excites
- L24:1015 excites
- L21:2044 excites
- L20:2489 excites
- L16:1384 excites
- L19:2333 excites
- L19:759 excites
- L17:2422 excites
- L22:408 excites
- L13:2796 excites
- L14:1171 excites
- L20:1722 excites
- L20:409 suppresses
- L12:1909 weakly excites
- L15:2098 weakly excites
- L17:1437 weakly excites
- L20:1411 weakly excites
- L20:1617 weakly excites
- L23:2134 weakly excites
- L24:2293 weakly excites
- L16:1367 weakly excites
- L25:1627 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
- supply
- _supply
- 供应
- Supply
- supply
- shortages
- mon
- supplies
- ius
- sup
Suppresses
- erotica
- TRIES
- aptcha
- endforeach
- etros
- tement
- ceremonial
- 欧式
- Tart
- voyeur