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

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

Neuron 2730 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.4072. Min Act: -0.0000

Data Index: 460355 (C4 (Web Text))

Max Activating Token Index: 596

Text #1

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.3670. Min Act: -0.0000

Data Index: 269365 (C4 (Web Text))

Max Activating Token Index: 230

Text #2

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.3418. Min Act: -0.0000

Data Index: 741112 (C4 (Web Text))

Max Activating Token Index: 908

Text #3

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.3362. Min Act: -0.0000

Data Index: 948088 (C4 (Web Text))

Max Activating Token Index: 171

Text #4

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.3078. Min Act: -0.0001

Data Index: 1030804 (C4 (Web Text))

Max Activating Token Index: 629

Text #5

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.3174. Min Act: -0.0001

Data Index: 421836 (C4 (Web Text))

Max Activating Token Index: 761

Text #6

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2909. Min Act: -0.0001

Data Index: 627215 (C4 (Web Text))

Max Activating Token Index: 22

Text #7

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.3056. Min Act: -0.0000

Data Index: 569480 (C4 (Web Text))

Max Activating Token Index: 100

Text #8

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2937. Min Act: -0.0001

Data Index: 1105502 (C4 (Web Text))

Max Activating Token Index: 753

Text #9

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2805. Min Act: -0.0000

Data Index: 1359491 (C4 (Web Text))

Max Activating Token Index: 463

Text #10

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2986. Min Act: -0.0000

Data Index: 534599 (C4 (Web Text))

Max Activating Token Index: 524

Text #11

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2878. Min Act: -0.0001

Data Index: 206046 (C4 (Web Text))

Max Activating Token Index: 508

Text #12

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2890. Min Act: -0.0001

Data Index: 715281 (C4 (Web Text))

Max Activating Token Index: 111

Text #13

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2752. Min Act: -0.0001

Data Index: 1165132 (C4 (Web Text))

Max Activating Token Index: 908

Text #14

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2843. Min Act: -0.0001

Data Index: 856624 (C4 (Web Text))

Max Activating Token Index: 797

Text #15

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2721. Min Act: -0.0000

Data Index: 945136 (C4 (Web Text))

Max Activating Token Index: 422

Text #16

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2821. Min Act: -0.0000

Data Index: 1186330 (C4 (Web Text))

Max Activating Token Index: 711

Text #17

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2712. Min Act: -0.0001

Data Index: 1367570 (Python Code)

Max Activating Token Index: 462

Text #18

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2890. Min Act: -0.0001

Data Index: 702527 (C4 (Web Text))

Max Activating Token Index: 210

Text #19

Max Range: 0.4072. Min Range: -0.4072

Max Act: 0.2783. Min Act: -0.0001

Data Index: 661160 (C4 (Web Text))

Max Activating Token Index: 254