Model: GELU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 1986 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.3434. Min Act: -0.1700

Data Index: 684471 (C4 (Web Text))

Max Activating Token Index: 190

Text #1

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.4059. Min Act: -0.1700

Data Index: 121361 (C4 (Web Text))

Max Activating Token Index: 751

Text #2

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.3746. Min Act: -0.1700

Data Index: 1040384 (C4 (Web Text))

Max Activating Token Index: 551

Text #3

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.2496. Min Act: -0.1700

Data Index: 573444 (C4 (Web Text))

Max Activating Token Index: 899

Text #4

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.1871. Min Act: -0.1700

Data Index: 604999 (C4 (Web Text))

Max Activating Token Index: 906

Text #5

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.1559. Min Act: -0.1700

Data Index: 390568 (C4 (Web Text))

Max Activating Token Index: 253

Text #6

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.1559. Min Act: -0.1700

Data Index: 893253 (C4 (Web Text))

Max Activating Token Index: 199

Text #7

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.1559. Min Act: -0.1700

Data Index: 408262 (C4 (Web Text))

Max Activating Token Index: 82

Text #8

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.1246. Min Act: -0.1700

Data Index: 365136 (C4 (Web Text))

Max Activating Token Index: 293

Text #9

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.0934. Min Act: -0.1700

Data Index: 1123464 (C4 (Web Text))

Max Activating Token Index: 493

Text #10

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.1246. Min Act: -0.1700

Data Index: 1190641 (C4 (Web Text))

Max Activating Token Index: 79

Text #11

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.1246. Min Act: -0.1700

Data Index: 310209 (C4 (Web Text))

Max Activating Token Index: 743

Text #12

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.0934. Min Act: -0.1700

Data Index: 513698 (C4 (Web Text))

Max Activating Token Index: 1001

Text #13

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.0621. Min Act: -0.1700

Data Index: 376075 (C4 (Web Text))

Max Activating Token Index: 263

Text #14

Max Range: 5.4059. Min Range: -5.4059

Max Act: 4.9996. Min Act: -0.1700

Data Index: 1228623 (C4 (Web Text))

Max Activating Token Index: 327

Text #15

Max Range: 5.4059. Min Range: -5.4059

Max Act: 5.0309. Min Act: -0.1700

Data Index: 1240897 (C4 (Web Text))

Max Activating Token Index: 434

Text #16

Max Range: 5.4059. Min Range: -5.4059

Max Act: 4.9684. Min Act: -0.1700

Data Index: 346269 (C4 (Web Text))

Max Activating Token Index: 872

Text #17

Max Range: 5.4059. Min Range: -5.4059

Max Act: 4.9996. Min Act: -0.1700

Data Index: 1025610 (C4 (Web Text))

Max Activating Token Index: 496

Text #18

Max Range: 5.4059. Min Range: -5.4059

Max Act: 4.9996. Min Act: -0.1700

Data Index: 323711 (C4 (Web Text))

Max Activating Token Index: 870

Text #19

Max Range: 5.4059. Min Range: -5.4059

Max Act: 4.9684. Min Act: -0.1700

Data Index: 64075 (C4 (Web Text))

Max Activating Token Index: 355