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

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

Neuron 4902 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.3114. Min Range: -1.3114

Max Act: 1.3114. Min Act: -0.0001

Data Index: 571970 (C4 (Web Text))

Max Activating Token Index: 105

Text #1

Max Range: 1.3114. Min Range: -1.3114

Max Act: 1.2446. Min Act: -0.0001

Data Index: 144132 (C4 (Web Text))

Max Activating Token Index: 464

Text #2

Max Range: 1.3114. Min Range: -1.3114

Max Act: 1.0832. Min Act: -0.0001

Data Index: 870494 (C4 (Web Text))

Max Activating Token Index: 682

Text #3

Max Range: 1.3114. Min Range: -1.3114

Max Act: 1.1119. Min Act: -0.0001

Data Index: 980856 (C4 (Web Text))

Max Activating Token Index: 833

Text #4

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8975. Min Act: -0.0001

Data Index: 1151430 (C4 (Web Text))

Max Activating Token Index: 193

Text #5

Max Range: 1.3114. Min Range: -1.3114

Max Act: 1.0008. Min Act: -0.0001

Data Index: 924551 (C4 (Web Text))

Max Activating Token Index: 10

Text #6

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.9705. Min Act: -0.0001

Data Index: 546190 (C4 (Web Text))

Max Activating Token Index: 119

Text #7

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.9198. Min Act: -0.0001

Data Index: 159589 (C4 (Web Text))

Max Activating Token Index: 343

Text #8

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.7851. Min Act: -0.0001

Data Index: 1578133 (Python Code)

Max Activating Token Index: 986

Text #9

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.7979. Min Act: -0.0001

Data Index: 763566 (C4 (Web Text))

Max Activating Token Index: 911

Text #10

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.7922. Min Act: -0.0001

Data Index: 1479616 (Python Code)

Max Activating Token Index: 36

Text #11

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8941. Min Act: -0.0001

Data Index: 1520672 (Python Code)

Max Activating Token Index: 748

Text #12

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8380. Min Act: -0.0001

Data Index: 1256485 (C4 (Web Text))

Max Activating Token Index: 328

Text #13

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.7779. Min Act: -0.0001

Data Index: 275683 (C4 (Web Text))

Max Activating Token Index: 78

Text #14

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8130. Min Act: -0.0001

Data Index: 1520934 (Python Code)

Max Activating Token Index: 10

Text #15

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8319. Min Act: -0.0001

Data Index: 1322904 (C4 (Web Text))

Max Activating Token Index: 324

Text #16

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8246. Min Act: -0.0001

Data Index: 1572007 (Python Code)

Max Activating Token Index: 1019

Text #17

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.7970. Min Act: -0.0001

Data Index: 1605451 (Python Code)

Max Activating Token Index: 168

Text #18

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8270. Min Act: -0.0001

Data Index: 56962 (C4 (Web Text))

Max Activating Token Index: 7

Text #19

Max Range: 1.3114. Min Range: -1.3114

Max Act: 0.8840. Min Act: -0.0001

Data Index: 735601 (C4 (Web Text))

Max Activating Token Index: 115