Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

Dataset: 80% C4 (Web Text) and 20% Python Code

Neuron 5900 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.1531. Min Act: -0.0001

Data Index: 606257 (C4 (Web Text))

Max Activating Token Index: 373

Text #1

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.1202. Min Act: -0.0001

Data Index: 995061 (C4 (Web Text))

Max Activating Token Index: 554

Text #2

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.1087. Min Act: -0.0001

Data Index: 212240 (C4 (Web Text))

Max Activating Token Index: 327

Text #3

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.1004. Min Act: -0.0001

Data Index: 989561 (C4 (Web Text))

Max Activating Token Index: 230

Text #4

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0974. Min Act: -0.0001

Data Index: 824844 (C4 (Web Text))

Max Activating Token Index: 78

Text #5

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0805. Min Act: -0.0001

Data Index: 24844 (C4 (Web Text))

Max Activating Token Index: 67

Text #6

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0795. Min Act: -0.0001

Data Index: 969900 (C4 (Web Text))

Max Activating Token Index: 166

Text #7

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0910. Min Act: -0.0001

Data Index: 93173 (C4 (Web Text))

Max Activating Token Index: 194

Text #8

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0797. Min Act: -0.0001

Data Index: 25078 (C4 (Web Text))

Max Activating Token Index: 852

Text #9

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0806. Min Act: -0.0001

Data Index: 957785 (C4 (Web Text))

Max Activating Token Index: 789

Text #10

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0622. Min Act: -0.0001

Data Index: 761305 (C4 (Web Text))

Max Activating Token Index: 182

Text #11

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0668. Min Act: -0.0001

Data Index: 170029 (C4 (Web Text))

Max Activating Token Index: 220

Text #12

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0769. Min Act: -0.0001

Data Index: 1522521 (Python Code)

Max Activating Token Index: 187

Text #13

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0670. Min Act: -0.0001

Data Index: 22691 (C4 (Web Text))

Max Activating Token Index: 909

Text #14

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0760. Min Act: -0.0001

Data Index: 445723 (C4 (Web Text))

Max Activating Token Index: 903

Text #15

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0694. Min Act: -0.0001

Data Index: 370760 (C4 (Web Text))

Max Activating Token Index: 900

Text #16

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0655. Min Act: -0.0001

Data Index: 817200 (C4 (Web Text))

Max Activating Token Index: 33

Text #17

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0649. Min Act: -0.0001

Data Index: 659869 (C4 (Web Text))

Max Activating Token Index: 449

Text #18

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0675. Min Act: -0.0001

Data Index: 214434 (C4 (Web Text))

Max Activating Token Index: 4

Text #19

Max Range: 0.1531. Min Range: -0.1531

Max Act: 0.0628. Min Act: -0.0001

Data Index: 54496 (C4 (Web Text))

Max Activating Token Index: 227