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

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

Neuron 251 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.3070. Min Act: -0.0001

Data Index: 1252966 (C4 (Web Text))

Max Activating Token Index: 92

Text #1

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2970. Min Act: -0.0001

Data Index: 601523 (C4 (Web Text))

Max Activating Token Index: 917

Text #2

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.3089. Min Act: -0.0001

Data Index: 743608 (C4 (Web Text))

Max Activating Token Index: 978

Text #3

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2886. Min Act: -0.0001

Data Index: 122252 (C4 (Web Text))

Max Activating Token Index: 245

Text #4

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2882. Min Act: -0.0001

Data Index: 513896 (C4 (Web Text))

Max Activating Token Index: 968

Text #5

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.3014. Min Act: -0.0001

Data Index: 676519 (C4 (Web Text))

Max Activating Token Index: 88

Text #6

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.3007. Min Act: -0.0001

Data Index: 1213161 (C4 (Web Text))

Max Activating Token Index: 33

Text #7

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2856. Min Act: -0.0001

Data Index: 940304 (C4 (Web Text))

Max Activating Token Index: 279

Text #8

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2758. Min Act: -0.0001

Data Index: 359707 (C4 (Web Text))

Max Activating Token Index: 15

Text #9

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2593. Min Act: -0.0001

Data Index: 641371 (C4 (Web Text))

Max Activating Token Index: 32

Text #10

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2210. Min Act: -0.0001

Data Index: 147594 (C4 (Web Text))

Max Activating Token Index: 576

Text #11

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2596. Min Act: -0.0001

Data Index: 1639372 (Python Code)

Max Activating Token Index: 582

Text #12

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2406. Min Act: -0.0001

Data Index: 1401618 (Python Code)

Max Activating Token Index: 704

Text #13

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2478. Min Act: -0.0001

Data Index: 1483456 (Python Code)

Max Activating Token Index: 731

Text #14

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2559. Min Act: -0.0001

Data Index: 1607206 (Python Code)

Max Activating Token Index: 74

Text #15

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2480. Min Act: -0.0001

Data Index: 1582673 (Python Code)

Max Activating Token Index: 667

Text #16

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2476. Min Act: -0.0001

Data Index: 446991 (C4 (Web Text))

Max Activating Token Index: 504

Text #17

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2483. Min Act: -0.0001

Data Index: 136566 (C4 (Web Text))

Max Activating Token Index: 179

Text #18

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2565. Min Act: -0.0001

Data Index: 1044505 (C4 (Web Text))

Max Activating Token Index: 218

Text #19

Max Range: 0.3089. Min Range: -0.3089

Max Act: 0.2377. Min Act: -0.0001

Data Index: 221119 (C4 (Web Text))

Max Activating Token Index: 371