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

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

Neuron 1997 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.6087. Min Act: -0.1700

Data Index: 426680 (C4 (Web Text))

Max Activating Token Index: 883

Text #1

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5931. Min Act: -0.1700

Data Index: 883829 (C4 (Web Text))

Max Activating Token Index: 737

Text #2

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5461. Min Act: -0.1700

Data Index: 668302 (C4 (Web Text))

Max Activating Token Index: 307

Text #3

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5304. Min Act: -0.1700

Data Index: 876187 (C4 (Web Text))

Max Activating Token Index: 634

Text #4

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5147. Min Act: -0.1700

Data Index: 539619 (C4 (Web Text))

Max Activating Token Index: 169

Text #5

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5304. Min Act: -0.1700

Data Index: 1253903 (C4 (Web Text))

Max Activating Token Index: 692

Text #6

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5304. Min Act: -0.1700

Data Index: 1131020 (C4 (Web Text))

Max Activating Token Index: 492

Text #7

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5617. Min Act: -0.1700

Data Index: 1002758 (C4 (Web Text))

Max Activating Token Index: 274

Text #8

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4834. Min Act: -0.1700

Data Index: 1607072 (Python Code)

Max Activating Token Index: 528

Text #9

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.5147. Min Act: -0.1700

Data Index: 963984 (C4 (Web Text))

Max Activating Token Index: 709

Text #10

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4834. Min Act: -0.1700

Data Index: 723885 (C4 (Web Text))

Max Activating Token Index: 993

Text #11

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4677. Min Act: -0.1700

Data Index: 698713 (C4 (Web Text))

Max Activating Token Index: 767

Text #12

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4521. Min Act: -0.1700

Data Index: 1535490 (Python Code)

Max Activating Token Index: 349

Text #13

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4834. Min Act: -0.1700

Data Index: 1632692 (Python Code)

Max Activating Token Index: 381

Text #14

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4521. Min Act: -0.1700

Data Index: 1415042 (Python Code)

Max Activating Token Index: 237

Text #15

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4521. Min Act: -0.1700

Data Index: 1570119 (Python Code)

Max Activating Token Index: 569

Text #16

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4364. Min Act: -0.1700

Data Index: 942812 (C4 (Web Text))

Max Activating Token Index: 361

Text #17

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4364. Min Act: -0.1700

Data Index: 780718 (C4 (Web Text))

Max Activating Token Index: 508

Text #18

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4207. Min Act: -0.1700

Data Index: 1238013 (C4 (Web Text))

Max Activating Token Index: 835

Text #19

Max Range: 3.6087. Min Range: -3.6087

Max Act: 3.4364. Min Act: -0.1700

Data Index: 120041 (C4 (Web Text))

Max Activating Token Index: 863