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

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

Neuron 2019 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: 5.6351. Min Range: -5.6351

Max Act: 5.6351. Min Act: -0.1700

Data Index: 604842 (C4 (Web Text))

Max Activating Token Index: 556

Text #1

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.6351. Min Act: -0.1700

Data Index: 844125 (C4 (Web Text))

Max Activating Token Index: 429

Text #2

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.5414. Min Act: -0.1700

Data Index: 836375 (C4 (Web Text))

Max Activating Token Index: 260

Text #3

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.5258. Min Act: -0.1700

Data Index: 606168 (C4 (Web Text))

Max Activating Token Index: 464

Text #4

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.5414. Min Act: -0.1700

Data Index: 1015694 (C4 (Web Text))

Max Activating Token Index: 364

Text #5

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.4164. Min Act: -0.1700

Data Index: 905686 (C4 (Web Text))

Max Activating Token Index: 194

Text #6

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.4476. Min Act: -0.1700

Data Index: 830079 (C4 (Web Text))

Max Activating Token Index: 770

Text #7

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.4164. Min Act: -0.1700

Data Index: 971011 (C4 (Web Text))

Max Activating Token Index: 363

Text #8

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.4320. Min Act: -0.1700

Data Index: 576907 (C4 (Web Text))

Max Activating Token Index: 669

Text #9

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.4164. Min Act: -0.1700

Data Index: 1106924 (C4 (Web Text))

Max Activating Token Index: 549

Text #10

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.4008. Min Act: -0.1699

Data Index: 1012847 (C4 (Web Text))

Max Activating Token Index: 202

Text #11

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.4633. Min Act: -0.1700

Data Index: 724247 (C4 (Web Text))

Max Activating Token Index: 989

Text #12

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.3539. Min Act: -0.1700

Data Index: 595594 (C4 (Web Text))

Max Activating Token Index: 326

Text #13

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.3539. Min Act: -0.1699

Data Index: 1575058 (Python Code)

Max Activating Token Index: 1023

Text #14

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.3539. Min Act: -0.1700

Data Index: 862719 (C4 (Web Text))

Max Activating Token Index: 698

Text #15

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.3070. Min Act: -0.1700

Data Index: 1161207 (C4 (Web Text))

Max Activating Token Index: 267

Text #16

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.2758. Min Act: -0.1700

Data Index: 250518 (C4 (Web Text))

Max Activating Token Index: 835

Text #17

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.2914. Min Act: -0.1700

Data Index: 303606 (C4 (Web Text))

Max Activating Token Index: 930

Text #18

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.2758. Min Act: -0.1700

Data Index: 613354 (C4 (Web Text))

Max Activating Token Index: 926

Text #19

Max Range: 5.6351. Min Range: -5.6351

Max Act: 5.2758. Min Act: -0.1700

Data Index: 643094 (C4 (Web Text))

Max Activating Token Index: 473