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

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

Neuron 340 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: 5.6687. Min Range: -5.6687

Max Act: 5.6687. Min Act: -0.1700

Data Index: 1329107 (C4 (Web Text))

Max Activating Token Index: 943

Text #1

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.3249. Min Act: -0.1700

Data Index: 814600 (C4 (Web Text))

Max Activating Token Index: 764

Text #2

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.3249. Min Act: -0.1700

Data Index: 257680 (C4 (Web Text))

Max Activating Token Index: 945

Text #3

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.2937. Min Act: -0.1700

Data Index: 330917 (C4 (Web Text))

Max Activating Token Index: 617

Text #4

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.2624. Min Act: -0.1700

Data Index: 1374126 (Python Code)

Max Activating Token Index: 113

Text #5

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.2937. Min Act: -0.1700

Data Index: 431386 (C4 (Web Text))

Max Activating Token Index: 486

Text #6

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.2624. Min Act: -0.1700

Data Index: 1230609 (C4 (Web Text))

Max Activating Token Index: 779

Text #7

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.2312. Min Act: -0.1700

Data Index: 36298 (C4 (Web Text))

Max Activating Token Index: 225

Text #8

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1999. Min Act: -0.1700

Data Index: 1086502 (C4 (Web Text))

Max Activating Token Index: 306

Text #9

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.2312. Min Act: -0.1700

Data Index: 1479957 (Python Code)

Max Activating Token Index: 197

Text #10

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1374. Min Act: -0.1700

Data Index: 146020 (C4 (Web Text))

Max Activating Token Index: 996

Text #11

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1687. Min Act: -0.1700

Data Index: 972339 (C4 (Web Text))

Max Activating Token Index: 903

Text #12

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1687. Min Act: -0.1700

Data Index: 1020699 (C4 (Web Text))

Max Activating Token Index: 516

Text #13

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1687. Min Act: -0.1700

Data Index: 155407 (C4 (Web Text))

Max Activating Token Index: 200

Text #14

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.2312. Min Act: -0.1700

Data Index: 226149 (C4 (Web Text))

Max Activating Token Index: 801

Text #15

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1374. Min Act: -0.1700

Data Index: 1656078 (Python Code)

Max Activating Token Index: 138

Text #16

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1374. Min Act: -0.1700

Data Index: 445616 (C4 (Web Text))

Max Activating Token Index: 204

Text #17

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.0749. Min Act: -0.1700

Data Index: 706137 (C4 (Web Text))

Max Activating Token Index: 223

Text #18

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.0437. Min Act: -0.1700

Data Index: 447156 (C4 (Web Text))

Max Activating Token Index: 346

Text #19

Max Range: 5.6687. Min Range: -5.6687

Max Act: 5.1687. Min Act: -0.1700

Data Index: 601292 (C4 (Web Text))

Max Activating Token Index: 730