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

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

Neuron 1605 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.5060. Min Act: -0.1700

Data Index: 190441 (C4 (Web Text))

Max Activating Token Index: 696

Text #1

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.3649. Min Act: -0.1700

Data Index: 999808 (C4 (Web Text))

Max Activating Token Index: 879

Text #2

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.3492. Min Act: -0.1700

Data Index: 1291596 (C4 (Web Text))

Max Activating Token Index: 971

Text #3

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.3492. Min Act: -0.1700

Data Index: 1297498 (C4 (Web Text))

Max Activating Token Index: 758

Text #4

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2707. Min Act: -0.1700

Data Index: 527098 (C4 (Web Text))

Max Activating Token Index: 791

Text #5

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2707. Min Act: -0.1700

Data Index: 309394 (C4 (Web Text))

Max Activating Token Index: 736

Text #6

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2550. Min Act: -0.1700

Data Index: 1082038 (C4 (Web Text))

Max Activating Token Index: 720

Text #7

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2550. Min Act: -0.1700

Data Index: 1363146 (Python Code)

Max Activating Token Index: 477

Text #8

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2707. Min Act: -0.1700

Data Index: 1340149 (C4 (Web Text))

Max Activating Token Index: 856

Text #9

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2550. Min Act: -0.1700

Data Index: 1267794 (C4 (Web Text))

Max Activating Token Index: 535

Text #10

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2707. Min Act: -0.1700

Data Index: 988232 (C4 (Web Text))

Max Activating Token Index: 427

Text #11

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.1921. Min Act: -0.1700

Data Index: 1091702 (C4 (Web Text))

Max Activating Token Index: 679

Text #12

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2235. Min Act: -0.1700

Data Index: 1647051 (Python Code)

Max Activating Token Index: 144

Text #13

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.1921. Min Act: -0.1700

Data Index: 364026 (C4 (Web Text))

Max Activating Token Index: 179

Text #14

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.1763. Min Act: -0.1700

Data Index: 364435 (C4 (Web Text))

Max Activating Token Index: 453

Text #15

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2235. Min Act: -0.1700

Data Index: 1006410 (C4 (Web Text))

Max Activating Token Index: 469

Text #16

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2078. Min Act: -0.1700

Data Index: 237863 (C4 (Web Text))

Max Activating Token Index: 936

Text #17

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.1921. Min Act: -0.1700

Data Index: 1469622 (Python Code)

Max Activating Token Index: 124

Text #18

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.1921. Min Act: -0.1700

Data Index: 157511 (C4 (Web Text))

Max Activating Token Index: 239

Text #19

Max Range: 3.5060. Min Range: -3.5060

Max Act: 3.2078. Min Act: -0.1700

Data Index: 45060 (C4 (Web Text))

Max Activating Token Index: 496