"as if" / "as though" after seem or look
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. 22 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 "as" when it comes right after words like "seems", "looks" or "feels" and starts a comparison such as "as if" or "as though". When it was turned up, the model just kept writing the word "if" over and over.
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.
-
chairman Mahmoud Abbas is accusing Israel of “genocide,” and the PA’s newspaper is calling the war “Israel’s Holocaust.”⏎It seems as if every time Israel defends itself, somebody points an accusing finger and yells “
-
Frankenstein’s monster, consisting of two identical M1 Max chips bolted together using a silicon interface that serves as a bridge. This clever design makes it seem as if the conjoined chips are in fact just
-
y⏎May 27, 2012⏎When the last American astronauts blasted off from the Moon in 1972, it seemed as if they were leaving behind monuments that would stand for all time. On a life
-
(the form is at the bottom of the page).⏎Esther – Day 3⏎Xerxes wanted to show off his beautiful wife! But it seems as though he did not give her too much thought until he had drunk too much!
-
his book:⏎The two entities that were supposed to write the regulations, the Pay Board and the Price Commission, wrangled and dithered. When it looked as though they were going to miss a crucial deadline
-
because there is seldom just one. In the beginning, actinic keratoses are frequently so small that they are recognized by touch rather than sight. It feels as if you were running a finger over sandpaper. There are many times the number
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: how to make a way for the way to be seen, and how to make the way way to be seen in such a way that the way is seen, and the way is made in such a way that the way is seen. I
Most negative on
Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.
-
Mind and Life Institute dialogue. We highlight some of the passages Adele Diamond presented to the Dalai Lama in Dharamsala — including texts from Rabbi Heschel, Bashevis Singer, Rachel Naomi Remen, and Henri Nouwen
-
the cortical area just posterior to the auditory cortex (Heschl's gyrus) within the Sylvian fissure. It is a triangular
-
.⏎Plants were perfect and completely green and had a good amount<|endoftext|>A remarkable new species of tyrannosaur has been unearthed in Grand Staircase-Escalante National Monument (GSENM), southern Utah. The huge carnivore inhabited
-
. This class shows how such demographic dynamics influenced early jazz development and how the urban environment in promoted gender diversification and the social acceptance of jazz. Report on Hersch due. Read Lawrence Gushee, "A Preliminary Chron
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
- L20:3062 strongly suppresses
- L20:2859 suppresses
- L21:1695 suppresses
- L19:1815 suppresses
- L21:1592 suppresses
- L19:1332 excites
- L19:2478 suppresses
- L19:2769 excites
- L21:1064 suppresses
- L21:1705 excites
- L18:2163 excites
- L20:2572 excites
- L18:2901 suppresses
- L20:726 excites
- L20:2600 excites
- L18:2503 suppresses
- L19:1783 weakly excites
- L21:2847 weakly excites
- L19:3062 weakly excites
- L20:2967 weakly suppresses
- L1:524 weakly suppresses
- L8:852 weakly excites comma inside large numbers
- L18:2350 weakly excites
- L19:2916 weakly suppresses
Feeds later layers
- L25:768 strongly excites
- L27:511 strongly excites
- L23:449 strongly excites
- L26:509 strongly excites
- L24:1945 excites
- L27:1084 excites
- L24:100 weakly suppresses
- L27:3024 weakly excites
- L23:1475 weakly excites
- L23:1813 weakly suppresses
- L25:1387 weakly excites end-of-document token and code keywords
- L27:2056 weakly suppresses the 'w' in shipwreck
- L24:2085 weakly suppresses
- L23:2514 weakly suppresses
- L27:987 weakly suppresses
- L27:2696 weakly suppresses
- L26:265 weakly excites
- L24:1002 weakly suppresses
- L25:1238 weakly suppresses
- L25:2741 weakly suppresses
- L26:2640 weakly suppresses
- L27:29 weakly excites
- L27:2325 weakly suppresses
- L25:2776 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
- if
- 如果
- if
- 若
- If
- If
- →if
- если
- nếu
- 如果有
Suppresses
- DCALL
- 背上
- 회의
- 那么简单
- 并对
- agnar
- шин
- icc
- 意义上
- ared