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

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

Neuron 1686 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1534. Min Act: -0.0001

Data Index: 1131258 (C4 (Web Text))

Max Activating Token Index: 61

Text #1

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1285. Min Act: -0.0001

Data Index: 1208600 (C4 (Web Text))

Max Activating Token Index: 579

Text #2

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1168. Min Act: -0.0001

Data Index: 926783 (C4 (Web Text))

Max Activating Token Index: 563

Text #3

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1188. Min Act: -0.0001

Data Index: 671334 (C4 (Web Text))

Max Activating Token Index: 599

Text #4

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1194. Min Act: -0.0001

Data Index: 654921 (C4 (Web Text))

Max Activating Token Index: 895

Text #5

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1152. Min Act: -0.0002

Data Index: 87129 (C4 (Web Text))

Max Activating Token Index: 641

Text #6

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1177. Min Act: -0.0001

Data Index: 760096 (C4 (Web Text))

Max Activating Token Index: 11

Text #7

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1117. Min Act: -0.0001

Data Index: 73995 (C4 (Web Text))

Max Activating Token Index: 988

Text #8

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1087. Min Act: -0.0002

Data Index: 196199 (C4 (Web Text))

Max Activating Token Index: 159

Text #9

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1034. Min Act: -0.0001

Data Index: 1377498 (Python Code)

Max Activating Token Index: 349

Text #10

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1078. Min Act: -0.0002

Data Index: 385872 (C4 (Web Text))

Max Activating Token Index: 760

Text #11

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1048. Min Act: -0.0001

Data Index: 622898 (C4 (Web Text))

Max Activating Token Index: 646

Text #12

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1058. Min Act: -0.0002

Data Index: 643822 (C4 (Web Text))

Max Activating Token Index: 621

Text #13

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1073. Min Act: -0.0002

Data Index: 321929 (C4 (Web Text))

Max Activating Token Index: 527

Text #14

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1029. Min Act: -0.0001

Data Index: 1559056 (Python Code)

Max Activating Token Index: 596

Text #15

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1064. Min Act: -0.0002

Data Index: 1346667 (C4 (Web Text))

Max Activating Token Index: 201

Text #16

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.0950. Min Act: -0.0001

Data Index: 1343574 (C4 (Web Text))

Max Activating Token Index: 262

Text #17

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.1056. Min Act: -0.0002

Data Index: 497718 (C4 (Web Text))

Max Activating Token Index: 646

Text #18

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.0990. Min Act: -0.0001

Data Index: 609661 (C4 (Web Text))

Max Activating Token Index: 466

Text #19

Max Range: 0.1534. Min Range: -0.1534

Max Act: 0.0956. Min Act: -0.0001

Data Index: 59239 (C4 (Web Text))

Max Activating Token Index: 443