Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 3855 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 2.3173. Min Range: -2.3173

Max Act: 2.3173. Min Act: -0.0000

Data Index: 1577365 (Python Code)

Max Activating Token Index: 313

Text #1

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.6798. Min Act: -0.0000

Data Index: 509913 (C4 (Web Text))

Max Activating Token Index: 238

Text #2

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7616. Min Act: -0.0000

Data Index: 1182676 (C4 (Web Text))

Max Activating Token Index: 323

Text #3

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.8501. Min Act: -0.0000

Data Index: 80400 (C4 (Web Text))

Max Activating Token Index: 209

Text #4

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.9531. Min Act: -0.0000

Data Index: 1654 (C4 (Web Text))

Max Activating Token Index: 402

Text #5

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7752. Min Act: -0.0000

Data Index: 732010 (C4 (Web Text))

Max Activating Token Index: 77

Text #6

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7658. Min Act: -0.0000

Data Index: 224013 (C4 (Web Text))

Max Activating Token Index: 85

Text #7

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7482. Min Act: -0.0000

Data Index: 561062 (C4 (Web Text))

Max Activating Token Index: 398

Text #8

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7515. Min Act: -0.0000

Data Index: 1031663 (C4 (Web Text))

Max Activating Token Index: 101

Text #9

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7391. Min Act: -0.0000

Data Index: 1261040 (C4 (Web Text))

Max Activating Token Index: 290

Text #10

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7386. Min Act: -0.0000

Data Index: 328241 (C4 (Web Text))

Max Activating Token Index: 59

Text #11

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7060. Min Act: -0.0000

Data Index: 1392828 (Python Code)

Max Activating Token Index: 42

Text #12

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.6185. Min Act: -0.0000

Data Index: 719842 (C4 (Web Text))

Max Activating Token Index: 8

Text #13

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.6012. Min Act: -0.0000

Data Index: 1462914 (Python Code)

Max Activating Token Index: 443

Text #14

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7937. Min Act: -0.0000

Data Index: 323949 (C4 (Web Text))

Max Activating Token Index: 11

Text #15

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7716. Min Act: -0.0000

Data Index: 1074539 (C4 (Web Text))

Max Activating Token Index: 252

Text #16

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.6754. Min Act: -0.0000

Data Index: 455742 (C4 (Web Text))

Max Activating Token Index: 60

Text #17

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7714. Min Act: -0.0000

Data Index: 414962 (C4 (Web Text))

Max Activating Token Index: 23

Text #18

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7687. Min Act: -0.0000

Data Index: 1115539 (C4 (Web Text))

Max Activating Token Index: 80

Text #19

Max Range: 2.3173. Min Range: -2.3173

Max Act: 1.7541. Min Act: -0.0000

Data Index: 419919 (C4 (Web Text))

Max Activating Token Index: 239