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

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

Neuron 5069 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: 0.4892. Min Range: -0.4892

Max Act: 0.4892. Min Act: -0.0000

Data Index: 1584111 (Python Code)

Max Activating Token Index: 35

Text #1

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4836. Min Act: -0.0000

Data Index: 1227700 (C4 (Web Text))

Max Activating Token Index: 582

Text #2

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4770. Min Act: -0.0000

Data Index: 349778 (C4 (Web Text))

Max Activating Token Index: 360

Text #3

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4444. Min Act: -0.0000

Data Index: 763288 (C4 (Web Text))

Max Activating Token Index: 533

Text #4

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4403. Min Act: -0.0000

Data Index: 875633 (C4 (Web Text))

Max Activating Token Index: 112

Text #5

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4154. Min Act: -0.0000

Data Index: 482769 (C4 (Web Text))

Max Activating Token Index: 964

Text #6

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4365. Min Act: -0.0000

Data Index: 980779 (C4 (Web Text))

Max Activating Token Index: 948

Text #7

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4189. Min Act: -0.0000

Data Index: 322244 (C4 (Web Text))

Max Activating Token Index: 841

Text #8

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4092. Min Act: -0.0000

Data Index: 944280 (C4 (Web Text))

Max Activating Token Index: 680

Text #9

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4372. Min Act: -0.0000

Data Index: 1405638 (Python Code)

Max Activating Token Index: 117

Text #10

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4559. Min Act: -0.0000

Data Index: 653380 (C4 (Web Text))

Max Activating Token Index: 247

Text #11

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4348. Min Act: -0.0000

Data Index: 61952 (C4 (Web Text))

Max Activating Token Index: 360

Text #12

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.3860. Min Act: -0.0000

Data Index: 1463670 (Python Code)

Max Activating Token Index: 257

Text #13

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.3987. Min Act: -0.0000

Data Index: 1371758 (Python Code)

Max Activating Token Index: 123

Text #14

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.3836. Min Act: -0.0000

Data Index: 1434108 (Python Code)

Max Activating Token Index: 121

Text #15

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.3691. Min Act: -0.0000

Data Index: 434328 (C4 (Web Text))

Max Activating Token Index: 92

Text #16

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.4048. Min Act: -0.0000

Data Index: 1517161 (Python Code)

Max Activating Token Index: 346

Text #17

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.3737. Min Act: -0.0000

Data Index: 1472596 (Python Code)

Max Activating Token Index: 305

Text #18

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.3847. Min Act: -0.0000

Data Index: 1543630 (Python Code)

Max Activating Token Index: 48

Text #19

Max Range: 0.4892. Min Range: -0.4892

Max Act: 0.3469. Min Act: -0.0000

Data Index: 1338064 (C4 (Web Text))

Max Activating Token Index: 559