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

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

Neuron 1512 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.3899. Min Act: -0.0001

Data Index: 879116 (C4 (Web Text))

Max Activating Token Index: 293

Text #1

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2678. Min Act: -0.0001

Data Index: 141427 (C4 (Web Text))

Max Activating Token Index: 113

Text #2

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2633. Min Act: -0.0001

Data Index: 595458 (C4 (Web Text))

Max Activating Token Index: 91

Text #3

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2502. Min Act: -0.0001

Data Index: 711967 (C4 (Web Text))

Max Activating Token Index: 964

Text #4

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2430. Min Act: -0.0001

Data Index: 1258786 (C4 (Web Text))

Max Activating Token Index: 1018

Text #5

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2339. Min Act: -0.0001

Data Index: 1318296 (C4 (Web Text))

Max Activating Token Index: 324

Text #6

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2158. Min Act: -0.0001

Data Index: 1314089 (C4 (Web Text))

Max Activating Token Index: 221

Text #7

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2229. Min Act: -0.0001

Data Index: 558011 (C4 (Web Text))

Max Activating Token Index: 645

Text #8

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2062. Min Act: -0.0001

Data Index: 796004 (C4 (Web Text))

Max Activating Token Index: 548

Text #9

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1924. Min Act: -0.0001

Data Index: 449939 (C4 (Web Text))

Max Activating Token Index: 445

Text #10

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.2054. Min Act: -0.0001

Data Index: 1618647 (Python Code)

Max Activating Token Index: 449

Text #11

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1884. Min Act: -0.0001

Data Index: 1172820 (C4 (Web Text))

Max Activating Token Index: 208

Text #12

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1814. Min Act: -0.0001

Data Index: 1436660 (Python Code)

Max Activating Token Index: 500

Text #13

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1870. Min Act: -0.0001

Data Index: 173144 (C4 (Web Text))

Max Activating Token Index: 103

Text #14

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1851. Min Act: -0.0001

Data Index: 840870 (C4 (Web Text))

Max Activating Token Index: 918

Text #15

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1905. Min Act: -0.0001

Data Index: 724600 (C4 (Web Text))

Max Activating Token Index: 910

Text #16

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1828. Min Act: -0.0001

Data Index: 978482 (C4 (Web Text))

Max Activating Token Index: 498

Text #17

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1895. Min Act: -0.0001

Data Index: 1232014 (C4 (Web Text))

Max Activating Token Index: 96

Text #18

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1757. Min Act: -0.0001

Data Index: 374084 (C4 (Web Text))

Max Activating Token Index: 300

Text #19

Max Range: 0.3899. Min Range: -0.3899

Max Act: 0.1750. Min Act: -0.0001

Data Index: 1519671 (Python Code)

Max Activating Token Index: 247