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

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

Neuron 198 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: 0.8150. Min Range: -0.8150

Max Act: 0.8150. Min Act: -0.0001

Data Index: 1604510 (Python Code)

Max Activating Token Index: 177

Text #1

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.7388. Min Act: -0.0001

Data Index: 1056931 (C4 (Web Text))

Max Activating Token Index: 226

Text #2

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.8111. Min Act: -0.0001

Data Index: 229842 (C4 (Web Text))

Max Activating Token Index: 648

Text #3

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.7344. Min Act: -0.0001

Data Index: 205942 (C4 (Web Text))

Max Activating Token Index: 565

Text #4

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.6654. Min Act: -0.0001

Data Index: 731020 (C4 (Web Text))

Max Activating Token Index: 177

Text #5

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.6343. Min Act: -0.0001

Data Index: 1249141 (C4 (Web Text))

Max Activating Token Index: 204

Text #6

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.6258. Min Act: -0.0001

Data Index: 965928 (C4 (Web Text))

Max Activating Token Index: 314

Text #7

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5920. Min Act: -0.0001

Data Index: 1122760 (C4 (Web Text))

Max Activating Token Index: 303

Text #8

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.6181. Min Act: -0.0001

Data Index: 494545 (C4 (Web Text))

Max Activating Token Index: 103

Text #9

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.6560. Min Act: -0.0001

Data Index: 855647 (C4 (Web Text))

Max Activating Token Index: 224

Text #10

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5773. Min Act: -0.0001

Data Index: 940613 (C4 (Web Text))

Max Activating Token Index: 983

Text #11

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5777. Min Act: -0.0001

Data Index: 590485 (C4 (Web Text))

Max Activating Token Index: 543

Text #12

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5820. Min Act: -0.0001

Data Index: 822632 (C4 (Web Text))

Max Activating Token Index: 966

Text #13

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.6178. Min Act: -0.0001

Data Index: 532737 (C4 (Web Text))

Max Activating Token Index: 208

Text #14

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5901. Min Act: -0.0001

Data Index: 1037661 (C4 (Web Text))

Max Activating Token Index: 549

Text #15

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5889. Min Act: -0.0001

Data Index: 658660 (C4 (Web Text))

Max Activating Token Index: 115

Text #16

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5527. Min Act: -0.0001

Data Index: 221960 (C4 (Web Text))

Max Activating Token Index: 923

Text #17

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5910. Min Act: -0.0001

Data Index: 976129 (C4 (Web Text))

Max Activating Token Index: 236

Text #18

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5502. Min Act: -0.0001

Data Index: 1541146 (Python Code)

Max Activating Token Index: 549

Text #19

Max Range: 0.8150. Min Range: -0.8150

Max Act: 0.5855. Min Act: -0.0001

Data Index: 790988 (C4 (Web Text))

Max Activating Token Index: 119