Model: GELU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 1750 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.4531. Min Act: -0.1700

Data Index: 1511032 (Python Code)

Max Activating Token Index: 1023

Text #1

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.3433. Min Act: -0.1700

Data Index: 276832 (C4 (Web Text))

Max Activating Token Index: 454

Text #2

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.3590. Min Act: -0.1700

Data Index: 851763 (C4 (Web Text))

Max Activating Token Index: 275

Text #3

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.3433. Min Act: -0.1700

Data Index: 545113 (C4 (Web Text))

Max Activating Token Index: 887

Text #4

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.3433. Min Act: -0.1700

Data Index: 1561726 (Python Code)

Max Activating Token Index: 762

Text #5

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.3276. Min Act: -0.1700

Data Index: 203483 (C4 (Web Text))

Max Activating Token Index: 636

Text #6

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.3119. Min Act: -0.1700

Data Index: 1435058 (Python Code)

Max Activating Token Index: 1023

Text #7

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.3119. Min Act: -0.1700

Data Index: 967332 (C4 (Web Text))

Max Activating Token Index: 731

Text #8

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2962. Min Act: -0.1700

Data Index: 266914 (C4 (Web Text))

Max Activating Token Index: 303

Text #9

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2805. Min Act: -0.1700

Data Index: 32992 (C4 (Web Text))

Max Activating Token Index: 636

Text #10

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2962. Min Act: -0.1700

Data Index: 1139989 (C4 (Web Text))

Max Activating Token Index: 878

Text #11

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2962. Min Act: -0.1700

Data Index: 13541 (C4 (Web Text))

Max Activating Token Index: 469

Text #12

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2805. Min Act: -0.1700

Data Index: 1524792 (Python Code)

Max Activating Token Index: 935

Text #13

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2805. Min Act: -0.1700

Data Index: 1153058 (C4 (Web Text))

Max Activating Token Index: 149

Text #14

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2805. Min Act: -0.1700

Data Index: 536751 (C4 (Web Text))

Max Activating Token Index: 686

Text #15

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2805. Min Act: -0.1700

Data Index: 1020538 (C4 (Web Text))

Max Activating Token Index: 115

Text #16

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2805. Min Act: -0.1700

Data Index: 839070 (C4 (Web Text))

Max Activating Token Index: 298

Text #17

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2805. Min Act: -0.1700

Data Index: 1203175 (C4 (Web Text))

Max Activating Token Index: 986

Text #18

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2648. Min Act: -0.1700

Data Index: 512680 (C4 (Web Text))

Max Activating Token Index: 314

Text #19

Max Range: 3.4531. Min Range: -3.4531

Max Act: 3.2491. Min Act: -0.1700

Data Index: 1084360 (C4 (Web Text))

Max Activating Token Index: 788