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

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

Neuron 897 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: 7.4445. Min Range: -7.4445

Max Act: 7.4445. Min Act: -0.1700

Data Index: 1361923 (Python Code)

Max Activating Token Index: 952

Text #1

Max Range: 7.4445. Min Range: -7.4445

Max Act: 7.0695. Min Act: -0.1700

Data Index: 1224057 (C4 (Web Text))

Max Activating Token Index: 368

Text #2

Max Range: 7.4445. Min Range: -7.4445

Max Act: 7.0070. Min Act: -0.1700

Data Index: 754041 (C4 (Web Text))

Max Activating Token Index: 749

Text #3

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.8820. Min Act: -0.1700

Data Index: 1522147 (Python Code)

Max Activating Token Index: 587

Text #4

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.8976. Min Act: -0.1700

Data Index: 1261821 (C4 (Web Text))

Max Activating Token Index: 802

Text #5

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.8195. Min Act: -0.1700

Data Index: 1133808 (C4 (Web Text))

Max Activating Token Index: 281

Text #6

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.8664. Min Act: -0.1700

Data Index: 1577391 (Python Code)

Max Activating Token Index: 870

Text #7

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.7883. Min Act: -0.1700

Data Index: 621260 (C4 (Web Text))

Max Activating Token Index: 317

Text #8

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.7726. Min Act: -0.1700

Data Index: 1273031 (C4 (Web Text))

Max Activating Token Index: 213

Text #9

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.6476. Min Act: -0.1700

Data Index: 989555 (C4 (Web Text))

Max Activating Token Index: 994

Text #10

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.7101. Min Act: -0.1700

Data Index: 1284259 (C4 (Web Text))

Max Activating Token Index: 710

Text #11

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.7101. Min Act: -0.1700

Data Index: 5161 (C4 (Web Text))

Max Activating Token Index: 345

Text #12

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.6476. Min Act: -0.1700

Data Index: 916156 (C4 (Web Text))

Max Activating Token Index: 392

Text #13

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.7101. Min Act: -0.1700

Data Index: 967060 (C4 (Web Text))

Max Activating Token Index: 627

Text #14

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.6789. Min Act: -0.1700

Data Index: 380431 (C4 (Web Text))

Max Activating Token Index: 883

Text #15

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.6008. Min Act: -0.1700

Data Index: 753638 (C4 (Web Text))

Max Activating Token Index: 735

Text #16

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.7101. Min Act: -0.1700

Data Index: 721920 (C4 (Web Text))

Max Activating Token Index: 901

Text #17

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.6164. Min Act: -0.1700

Data Index: 266076 (C4 (Web Text))

Max Activating Token Index: 861

Text #18

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.7258. Min Act: -0.1700

Data Index: 1504990 (Python Code)

Max Activating Token Index: 795

Text #19

Max Range: 7.4445. Min Range: -7.4445

Max Act: 6.6945. Min Act: -0.1700

Data Index: 1407365 (Python Code)

Max Activating Token Index: 924