Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 3308 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.3231. Min Act: -0.0000

Data Index: 1003230 (C4 (Web Text))

Max Activating Token Index: 369

Text #1

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2694. Min Act: -0.0000

Data Index: 1097754 (C4 (Web Text))

Max Activating Token Index: 386

Text #2

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2473. Min Act: -0.0000

Data Index: 507028 (C4 (Web Text))

Max Activating Token Index: 360

Text #3

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2398. Min Act: -0.0000

Data Index: 947725 (C4 (Web Text))

Max Activating Token Index: 812

Text #4

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2439. Min Act: -0.0000

Data Index: 896097 (C4 (Web Text))

Max Activating Token Index: 758

Text #5

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2405. Min Act: -0.0000

Data Index: 957633 (C4 (Web Text))

Max Activating Token Index: 502

Text #6

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2521. Min Act: -0.0000

Data Index: 1389835 (Python Code)

Max Activating Token Index: 66

Text #7

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2125. Min Act: -0.0000

Data Index: 148640 (C4 (Web Text))

Max Activating Token Index: 164

Text #8

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2168. Min Act: 0.0000

Data Index: 1063836 (C4 (Web Text))

Max Activating Token Index: 510

Text #9

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.1983. Min Act: -0.0000

Data Index: 1361633 (Python Code)

Max Activating Token Index: 312

Text #10

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.1936. Min Act: -0.0000

Data Index: 816036 (C4 (Web Text))

Max Activating Token Index: 899

Text #11

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2060. Min Act: -0.0000

Data Index: 358091 (C4 (Web Text))

Max Activating Token Index: 931

Text #12

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.1987. Min Act: -0.0000

Data Index: 444950 (C4 (Web Text))

Max Activating Token Index: 779

Text #13

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2041. Min Act: -0.0000

Data Index: 217507 (C4 (Web Text))

Max Activating Token Index: 187

Text #14

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.1948. Min Act: -0.0000

Data Index: 1041515 (C4 (Web Text))

Max Activating Token Index: 278

Text #15

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.1955. Min Act: -0.0000

Data Index: 936947 (C4 (Web Text))

Max Activating Token Index: 862

Text #16

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2151. Min Act: -0.0000

Data Index: 1466787 (Python Code)

Max Activating Token Index: 144

Text #17

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.1963. Min Act: -0.0000

Data Index: 272008 (C4 (Web Text))

Max Activating Token Index: 157

Text #18

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.2017. Min Act: -0.0000

Data Index: 280665 (C4 (Web Text))

Max Activating Token Index: 915

Text #19

Max Range: 0.3231. Min Range: -0.3231

Max Act: 0.1837. Min Act: -0.0000

Data Index: 164771 (C4 (Web Text))

Max Activating Token Index: 268