Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 36 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.4142. Min Act: -0.0001

Data Index: 1033342 (C4 (Web Text))

Max Activating Token Index: 19

Text #1

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3768. Min Act: -0.0001

Data Index: 1399907 (Python Code)

Max Activating Token Index: 534

Text #2

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3767. Min Act: -0.0001

Data Index: 115042 (C4 (Web Text))

Max Activating Token Index: 696

Text #3

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3749. Min Act: -0.0001

Data Index: 218243 (C4 (Web Text))

Max Activating Token Index: 82

Text #4

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3703. Min Act: -0.0000

Data Index: 752605 (C4 (Web Text))

Max Activating Token Index: 420

Text #5

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3577. Min Act: -0.0001

Data Index: 1234457 (C4 (Web Text))

Max Activating Token Index: 300

Text #6

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3576. Min Act: -0.0001

Data Index: 1336707 (C4 (Web Text))

Max Activating Token Index: 396

Text #7

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3461. Min Act: -0.0001

Data Index: 1618309 (Python Code)

Max Activating Token Index: 391

Text #8

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3579. Min Act: -0.0001

Data Index: 554516 (C4 (Web Text))

Max Activating Token Index: 53

Text #9

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3499. Min Act: -0.0001

Data Index: 1416149 (Python Code)

Max Activating Token Index: 57

Text #10

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3450. Min Act: -0.0001

Data Index: 1494603 (Python Code)

Max Activating Token Index: 19

Text #11

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3418. Min Act: -0.0001

Data Index: 932254 (C4 (Web Text))

Max Activating Token Index: 1010

Text #12

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3626. Min Act: -0.0001

Data Index: 363383 (C4 (Web Text))

Max Activating Token Index: 635

Text #13

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3390. Min Act: -0.0001

Data Index: 1399371 (Python Code)

Max Activating Token Index: 385

Text #14

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3288. Min Act: -0.0001

Data Index: 1560418 (Python Code)

Max Activating Token Index: 317

Text #15

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3491. Min Act: -0.0001

Data Index: 1091255 (C4 (Web Text))

Max Activating Token Index: 525

Text #16

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3246. Min Act: -0.0001

Data Index: 565028 (C4 (Web Text))

Max Activating Token Index: 130

Text #17

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3182. Min Act: -0.0000

Data Index: 1042015 (C4 (Web Text))

Max Activating Token Index: 527

Text #18

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3155. Min Act: -0.0001

Data Index: 446743 (C4 (Web Text))

Max Activating Token Index: 712

Text #19

Max Range: 0.4142. Min Range: -0.4142

Max Act: 0.3363. Min Act: -0.0000

Data Index: 1174744 (C4 (Web Text))

Max Activating Token Index: 231