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

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

Neuron 2019 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.8270. Min Act: -0.1700

Data Index: 898519 (C4 (Web Text))

Max Activating Token Index: 962

Text #1

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.6549. Min Act: -0.1700

Data Index: 49044 (C4 (Web Text))

Max Activating Token Index: 875

Text #2

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.6392. Min Act: -0.1700

Data Index: 9093 (C4 (Web Text))

Max Activating Token Index: 776

Text #3

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.6079. Min Act: -0.1700

Data Index: 257659 (C4 (Web Text))

Max Activating Token Index: 654

Text #4

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.5922. Min Act: -0.1700

Data Index: 1119424 (C4 (Web Text))

Max Activating Token Index: 721

Text #5

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.5609. Min Act: -0.1700

Data Index: 613137 (C4 (Web Text))

Max Activating Token Index: 540

Text #6

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.5453. Min Act: -0.1700

Data Index: 1119134 (C4 (Web Text))

Max Activating Token Index: 867

Text #7

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.5453. Min Act: -0.1700

Data Index: 878602 (C4 (Web Text))

Max Activating Token Index: 64

Text #8

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.5296. Min Act: -0.1700

Data Index: 211174 (C4 (Web Text))

Max Activating Token Index: 893

Text #9

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.5453. Min Act: -0.1700

Data Index: 697999 (C4 (Web Text))

Max Activating Token Index: 181

Text #10

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.4826. Min Act: -0.1700

Data Index: 627431 (C4 (Web Text))

Max Activating Token Index: 898

Text #11

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.4356. Min Act: -0.1700

Data Index: 727231 (C4 (Web Text))

Max Activating Token Index: 939

Text #12

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3885. Min Act: -0.1700

Data Index: 1104396 (C4 (Web Text))

Max Activating Token Index: 78

Text #13

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3728. Min Act: -0.1700

Data Index: 779842 (C4 (Web Text))

Max Activating Token Index: 974

Text #14

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3728. Min Act: -0.1700

Data Index: 1466644 (Python Code)

Max Activating Token Index: 604

Text #15

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3414. Min Act: -0.1700

Data Index: 322537 (C4 (Web Text))

Max Activating Token Index: 297

Text #16

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3414. Min Act: -0.1700

Data Index: 749246 (C4 (Web Text))

Max Activating Token Index: 439

Text #17

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3100. Min Act: -0.1700

Data Index: 98448 (C4 (Web Text))

Max Activating Token Index: 448

Text #18

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3100. Min Act: -0.1700

Data Index: 325029 (C4 (Web Text))

Max Activating Token Index: 849

Text #19

Max Range: 3.8270. Min Range: -3.8270

Max Act: 3.3100. Min Act: -0.1700

Data Index: 308170 (C4 (Web Text))

Max Activating Token Index: 315