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

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

Neuron 3087 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: 0.2584. Min Range: -0.2584

Max Act: 0.2576. Min Act: -0.0001

Data Index: 984807 (C4 (Web Text))

Max Activating Token Index: 820

Text #1

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.2584. Min Act: -0.0001

Data Index: 299169 (C4 (Web Text))

Max Activating Token Index: 295

Text #2

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.2141. Min Act: -0.0001

Data Index: 757797 (C4 (Web Text))

Max Activating Token Index: 396

Text #3

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.2164. Min Act: -0.0000

Data Index: 649739 (C4 (Web Text))

Max Activating Token Index: 363

Text #4

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1984. Min Act: -0.0001

Data Index: 406341 (C4 (Web Text))

Max Activating Token Index: 941

Text #5

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.2112. Min Act: -0.0001

Data Index: 1494996 (Python Code)

Max Activating Token Index: 148

Text #6

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1971. Min Act: -0.0001

Data Index: 531725 (C4 (Web Text))

Max Activating Token Index: 739

Text #7

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1857. Min Act: -0.0001

Data Index: 962665 (C4 (Web Text))

Max Activating Token Index: 224

Text #8

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1802. Min Act: -0.0001

Data Index: 669820 (C4 (Web Text))

Max Activating Token Index: 1012

Text #9

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1778. Min Act: -0.0000

Data Index: 64521 (C4 (Web Text))

Max Activating Token Index: 671

Text #10

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1691. Min Act: -0.0001

Data Index: 98384 (C4 (Web Text))

Max Activating Token Index: 613

Text #11

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1865. Min Act: -0.0001

Data Index: 1588426 (Python Code)

Max Activating Token Index: 958

Text #12

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1943. Min Act: -0.0000

Data Index: 596190 (C4 (Web Text))

Max Activating Token Index: 661

Text #13

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1675. Min Act: -0.0000

Data Index: 1046720 (C4 (Web Text))

Max Activating Token Index: 244

Text #14

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1816. Min Act: -0.0000

Data Index: 562210 (C4 (Web Text))

Max Activating Token Index: 805

Text #15

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1839. Min Act: -0.0001

Data Index: 32190 (C4 (Web Text))

Max Activating Token Index: 155

Text #16

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1736. Min Act: -0.0000

Data Index: 565288 (C4 (Web Text))

Max Activating Token Index: 428

Text #17

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1635. Min Act: -0.0000

Data Index: 1544289 (Python Code)

Max Activating Token Index: 314

Text #18

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1909. Min Act: -0.0000

Data Index: 1170823 (C4 (Web Text))

Max Activating Token Index: 353

Text #19

Max Range: 0.2584. Min Range: -0.2584

Max Act: 0.1667. Min Act: -0.0000

Data Index: 951160 (C4 (Web Text))

Max Activating Token Index: 638