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

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

Neuron 192 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.9972. Min Act: -0.0001

Data Index: 1106257 (C4 (Web Text))

Max Activating Token Index: 951

Text #1

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.9250. Min Act: -0.0001

Data Index: 665920 (C4 (Web Text))

Max Activating Token Index: 8

Text #2

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.9344. Min Act: -0.0001

Data Index: 980182 (C4 (Web Text))

Max Activating Token Index: 19

Text #3

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.9478. Min Act: -0.0001

Data Index: 317823 (C4 (Web Text))

Max Activating Token Index: 18

Text #4

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8822. Min Act: -0.0001

Data Index: 256303 (C4 (Web Text))

Max Activating Token Index: 304

Text #5

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.9123. Min Act: -0.0001

Data Index: 363645 (C4 (Web Text))

Max Activating Token Index: 9

Text #6

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8802. Min Act: -0.0001

Data Index: 1317591 (C4 (Web Text))

Max Activating Token Index: 30

Text #7

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8482. Min Act: -0.0001

Data Index: 1553284 (Python Code)

Max Activating Token Index: 276

Text #8

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8597. Min Act: -0.0001

Data Index: 1512118 (Python Code)

Max Activating Token Index: 371

Text #9

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8277. Min Act: -0.0001

Data Index: 744609 (C4 (Web Text))

Max Activating Token Index: 69

Text #10

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8327. Min Act: -0.0001

Data Index: 1063360 (C4 (Web Text))

Max Activating Token Index: 376

Text #11

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8512. Min Act: -0.0001

Data Index: 790151 (C4 (Web Text))

Max Activating Token Index: 114

Text #12

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.7621. Min Act: -0.0001

Data Index: 569782 (C4 (Web Text))

Max Activating Token Index: 797

Text #13

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8102. Min Act: -0.0001

Data Index: 719005 (C4 (Web Text))

Max Activating Token Index: 40

Text #14

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8255. Min Act: -0.0001

Data Index: 629788 (C4 (Web Text))

Max Activating Token Index: 381

Text #15

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.7222. Min Act: -0.0001

Data Index: 1423805 (Python Code)

Max Activating Token Index: 850

Text #16

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.7588. Min Act: -0.0001

Data Index: 174113 (C4 (Web Text))

Max Activating Token Index: 153

Text #17

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.6950. Min Act: -0.0001

Data Index: 1232966 (C4 (Web Text))

Max Activating Token Index: 754

Text #18

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.8148. Min Act: -0.0001

Data Index: 303465 (C4 (Web Text))

Max Activating Token Index: 164

Text #19

Max Range: 0.9972. Min Range: -0.9972

Max Act: 0.7577. Min Act: -0.0001

Data Index: 538927 (C4 (Web Text))

Max Activating Token Index: 24