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

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

Neuron 3821 in Layer 10

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.5938. Min Act: 0.0000

Data Index: 64528 (C4 (Web Text))

Max Activating Token Index: 894

Text #1

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.5799. Min Act: -0.0000

Data Index: 144736 (C4 (Web Text))

Max Activating Token Index: 313

Text #2

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.5534. Min Act: -0.0000

Data Index: 968291 (C4 (Web Text))

Max Activating Token Index: 565

Text #3

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.5302. Min Act: -0.0000

Data Index: 921533 (C4 (Web Text))

Max Activating Token Index: 934

Text #4

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.5034. Min Act: -0.0000

Data Index: 1571262 (Python Code)

Max Activating Token Index: 256

Text #5

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.5204. Min Act: -0.0000

Data Index: 697017 (C4 (Web Text))

Max Activating Token Index: 842

Text #6

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.5147. Min Act: -0.0000

Data Index: 699883 (C4 (Web Text))

Max Activating Token Index: 280

Text #7

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4512. Min Act: -0.0000

Data Index: 123672 (C4 (Web Text))

Max Activating Token Index: 627

Text #8

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4603. Min Act: -0.0000

Data Index: 128108 (C4 (Web Text))

Max Activating Token Index: 698

Text #9

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4579. Min Act: -0.0000

Data Index: 1363814 (Python Code)

Max Activating Token Index: 278

Text #10

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4786. Min Act: -0.0000

Data Index: 164678 (C4 (Web Text))

Max Activating Token Index: 381

Text #11

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4564. Min Act: -0.0000

Data Index: 787883 (C4 (Web Text))

Max Activating Token Index: 496

Text #12

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4341. Min Act: -0.0000

Data Index: 1443070 (Python Code)

Max Activating Token Index: 26

Text #13

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4161. Min Act: -0.0000

Data Index: 1185008 (C4 (Web Text))

Max Activating Token Index: 962

Text #14

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4193. Min Act: -0.0000

Data Index: 1248889 (C4 (Web Text))

Max Activating Token Index: 392

Text #15

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4588. Min Act: -0.0000

Data Index: 230163 (C4 (Web Text))

Max Activating Token Index: 937

Text #16

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4101. Min Act: -0.0000

Data Index: 62806 (C4 (Web Text))

Max Activating Token Index: 261

Text #17

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.3979. Min Act: -0.0000

Data Index: 438752 (C4 (Web Text))

Max Activating Token Index: 80

Text #18

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.4095. Min Act: 0.0000

Data Index: 826953 (C4 (Web Text))

Max Activating Token Index: 752

Text #19

Max Range: 0.5938. Min Range: -0.5938

Max Act: 0.3909. Min Act: 0.0000

Data Index: 339797 (C4 (Web Text))

Max Activating Token Index: 985