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

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

Neuron 653 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: 1.3045. Min Range: -1.3045

Max Act: 1.3045. Min Act: -0.0001

Data Index: 573757 (C4 (Web Text))

Max Activating Token Index: 369

Text #1

Max Range: 1.3045. Min Range: -1.3045

Max Act: 1.2601. Min Act: -0.0001

Data Index: 956235 (C4 (Web Text))

Max Activating Token Index: 352

Text #2

Max Range: 1.3045. Min Range: -1.3045

Max Act: 1.2233. Min Act: -0.0001

Data Index: 84748 (C4 (Web Text))

Max Activating Token Index: 665

Text #3

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.9025. Min Act: -0.0001

Data Index: 911688 (C4 (Web Text))

Max Activating Token Index: 481

Text #4

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8665. Min Act: -0.0001

Data Index: 1568102 (Python Code)

Max Activating Token Index: 36

Text #5

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8628. Min Act: -0.0001

Data Index: 266183 (C4 (Web Text))

Max Activating Token Index: 579

Text #6

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8808. Min Act: -0.0001

Data Index: 546310 (C4 (Web Text))

Max Activating Token Index: 987

Text #7

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8405. Min Act: -0.0001

Data Index: 1478191 (Python Code)

Max Activating Token Index: 83

Text #8

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8087. Min Act: -0.0001

Data Index: 824896 (C4 (Web Text))

Max Activating Token Index: 219

Text #9

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8054. Min Act: -0.0001

Data Index: 679056 (C4 (Web Text))

Max Activating Token Index: 178

Text #10

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8594. Min Act: -0.0001

Data Index: 1435390 (Python Code)

Max Activating Token Index: 1016

Text #11

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7865. Min Act: -0.0001

Data Index: 370507 (C4 (Web Text))

Max Activating Token Index: 711

Text #12

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.8446. Min Act: -0.0001

Data Index: 429629 (C4 (Web Text))

Max Activating Token Index: 986

Text #13

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7857. Min Act: -0.0001

Data Index: 779800 (C4 (Web Text))

Max Activating Token Index: 336

Text #14

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7574. Min Act: -0.0001

Data Index: 1467954 (Python Code)

Max Activating Token Index: 313

Text #15

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7451. Min Act: -0.0001

Data Index: 1630856 (Python Code)

Max Activating Token Index: 911

Text #16

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7170. Min Act: -0.0001

Data Index: 812164 (C4 (Web Text))

Max Activating Token Index: 681

Text #17

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7619. Min Act: -0.0001

Data Index: 1517461 (Python Code)

Max Activating Token Index: 1019

Text #18

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7881. Min Act: -0.0001

Data Index: 7754 (C4 (Web Text))

Max Activating Token Index: 808

Text #19

Max Range: 1.3045. Min Range: -1.3045

Max Act: 0.7332. Min Act: -0.0001

Data Index: 247571 (C4 (Web Text))

Max Activating Token Index: 251