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

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

Neuron 931 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.9215. Min Act: -0.0001

Data Index: 592007 (C4 (Web Text))

Max Activating Token Index: 803

Text #1

Max Range: 1.0320. Min Range: -1.0320

Max Act: 1.0124. Min Act: -0.0001

Data Index: 701645 (C4 (Web Text))

Max Activating Token Index: 423

Text #2

Max Range: 1.0320. Min Range: -1.0320

Max Act: 1.0320. Min Act: -0.0001

Data Index: 623979 (C4 (Web Text))

Max Activating Token Index: 700

Text #3

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.9012. Min Act: -0.0001

Data Index: 936869 (C4 (Web Text))

Max Activating Token Index: 1022

Text #4

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8303. Min Act: -0.0001

Data Index: 456862 (C4 (Web Text))

Max Activating Token Index: 101

Text #5

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8967. Min Act: -0.0001

Data Index: 814274 (C4 (Web Text))

Max Activating Token Index: 21

Text #6

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8435. Min Act: -0.0001

Data Index: 631465 (C4 (Web Text))

Max Activating Token Index: 661

Text #7

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8629. Min Act: -0.0001

Data Index: 671750 (C4 (Web Text))

Max Activating Token Index: 611

Text #8

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8955. Min Act: -0.0001

Data Index: 984021 (C4 (Web Text))

Max Activating Token Index: 224

Text #9

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8947. Min Act: -0.0001

Data Index: 1619884 (Python Code)

Max Activating Token Index: 23

Text #10

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8671. Min Act: -0.0001

Data Index: 83153 (C4 (Web Text))

Max Activating Token Index: 804

Text #11

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.7984. Min Act: -0.0001

Data Index: 1136153 (C4 (Web Text))

Max Activating Token Index: 513

Text #12

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8085. Min Act: -0.0001

Data Index: 147364 (C4 (Web Text))

Max Activating Token Index: 340

Text #13

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8019. Min Act: -0.0001

Data Index: 1459536 (Python Code)

Max Activating Token Index: 113

Text #14

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.7931. Min Act: -0.0001

Data Index: 308865 (C4 (Web Text))

Max Activating Token Index: 142

Text #15

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.8169. Min Act: -0.0001

Data Index: 442297 (C4 (Web Text))

Max Activating Token Index: 936

Text #16

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.7371. Min Act: -0.0001

Data Index: 922849 (C4 (Web Text))

Max Activating Token Index: 669

Text #17

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.7480. Min Act: -0.0001

Data Index: 1311667 (C4 (Web Text))

Max Activating Token Index: 550

Text #18

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.7695. Min Act: -0.0001

Data Index: 233475 (C4 (Web Text))

Max Activating Token Index: 567

Text #19

Max Range: 1.0320. Min Range: -1.0320

Max Act: 0.7879. Min Act: -0.0001

Data Index: 248367 (C4 (Web Text))

Max Activating Token Index: 508