Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3174 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.9810. Min Act: -0.0001

Data Index: 1470811 (Python Code)

Max Activating Token Index: 470

Text #1

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.9482. Min Act: -0.0001

Data Index: 1415746 (Python Code)

Max Activating Token Index: 565

Text #2

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.8426. Min Act: -0.0001

Data Index: 1014817 (C4 (Web Text))

Max Activating Token Index: 66

Text #3

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.8550. Min Act: -0.0001

Data Index: 1483627 (Python Code)

Max Activating Token Index: 90

Text #4

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.8259. Min Act: -0.0001

Data Index: 1034438 (C4 (Web Text))

Max Activating Token Index: 108

Text #5

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7780. Min Act: -0.0001

Data Index: 1390089 (Python Code)

Max Activating Token Index: 739

Text #6

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.8104. Min Act: -0.0001

Data Index: 904351 (C4 (Web Text))

Max Activating Token Index: 662

Text #7

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7833. Min Act: -0.0001

Data Index: 1391954 (Python Code)

Max Activating Token Index: 360

Text #8

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7987. Min Act: -0.0001

Data Index: 1104200 (C4 (Web Text))

Max Activating Token Index: 684

Text #9

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7577. Min Act: -0.0001

Data Index: 1450369 (Python Code)

Max Activating Token Index: 983

Text #10

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7942. Min Act: -0.0001

Data Index: 1558176 (Python Code)

Max Activating Token Index: 912

Text #11

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7239. Min Act: -0.0001

Data Index: 1584379 (Python Code)

Max Activating Token Index: 675

Text #12

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7195. Min Act: -0.0001

Data Index: 1010336 (C4 (Web Text))

Max Activating Token Index: 15

Text #13

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7073. Min Act: -0.0001

Data Index: 1611775 (Python Code)

Max Activating Token Index: 273

Text #14

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7119. Min Act: -0.0001

Data Index: 1418348 (Python Code)

Max Activating Token Index: 278

Text #15

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.6781. Min Act: -0.0001

Data Index: 1549591 (Python Code)

Max Activating Token Index: 368

Text #16

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.6504. Min Act: -0.0001

Data Index: 1649795 (Python Code)

Max Activating Token Index: 905

Text #17

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7171. Min Act: -0.0001

Data Index: 1655725 (Python Code)

Max Activating Token Index: 579

Text #18

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.7422. Min Act: -0.0001

Data Index: 1588150 (Python Code)

Max Activating Token Index: 101

Text #19

Max Range: 0.9810. Min Range: -0.9810

Max Act: 0.6976. Min Act: -0.0001

Data Index: 1649176 (Python Code)

Max Activating Token Index: 726