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

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

Neuron 1983 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.3125. Min Act: -0.0000

Data Index: 790096 (C4 (Web Text))

Max Activating Token Index: 1002

Text #1

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.1523. Min Act: -0.0000

Data Index: 588379 (C4 (Web Text))

Max Activating Token Index: 964

Text #2

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.1896. Min Act: -0.0001

Data Index: 1443914 (Python Code)

Max Activating Token Index: 386

Text #3

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.1699. Min Act: -0.0001

Data Index: 794154 (C4 (Web Text))

Max Activating Token Index: 494

Text #4

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.1671. Min Act: -0.0001

Data Index: 321141 (C4 (Web Text))

Max Activating Token Index: 291

Text #5

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.2827. Min Act: -0.0000

Data Index: 255976 (C4 (Web Text))

Max Activating Token Index: 247

Text #6

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.0250. Min Act: -0.0000

Data Index: 1193872 (C4 (Web Text))

Max Activating Token Index: 249

Text #7

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.9572. Min Act: -0.0001

Data Index: 1545297 (Python Code)

Max Activating Token Index: 717

Text #8

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.8668. Min Act: -0.0000

Data Index: 1266561 (C4 (Web Text))

Max Activating Token Index: 621

Text #9

Max Range: 1.3125. Min Range: -1.3125

Max Act: 1.0055. Min Act: -0.0000

Data Index: 1622276 (Python Code)

Max Activating Token Index: 146

Text #10

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.9443. Min Act: -0.0001

Data Index: 1563069 (Python Code)

Max Activating Token Index: 802

Text #11

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.9262. Min Act: -0.0000

Data Index: 1364523 (Python Code)

Max Activating Token Index: 552

Text #12

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.9110. Min Act: -0.0001

Data Index: 1415916 (Python Code)

Max Activating Token Index: 262

Text #13

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.8857. Min Act: -0.0000

Data Index: 1088125 (C4 (Web Text))

Max Activating Token Index: 633

Text #14

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.8082. Min Act: -0.0001

Data Index: 1478166 (Python Code)

Max Activating Token Index: 413

Text #15

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.8490. Min Act: -0.0000

Data Index: 1035924 (C4 (Web Text))

Max Activating Token Index: 744

Text #16

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.8190. Min Act: -0.0001

Data Index: 1516501 (Python Code)

Max Activating Token Index: 79

Text #17

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.8145. Min Act: -0.0001

Data Index: 1512831 (Python Code)

Max Activating Token Index: 269

Text #18

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.7496. Min Act: -0.0001

Data Index: 1027372 (C4 (Web Text))

Max Activating Token Index: 159

Text #19

Max Range: 1.3125. Min Range: -1.3125

Max Act: 0.8208. Min Act: -0.0000

Data Index: 1100422 (C4 (Web Text))

Max Activating Token Index: 217