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

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

Neuron 18 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: 3.5742. Min Range: -3.5742

Max Act: 3.5742. Min Act: -0.1700

Data Index: 139269 (C4 (Web Text))

Max Activating Token Index: 314

Text #1

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.4489. Min Act: -0.1700

Data Index: 1551824 (Python Code)

Max Activating Token Index: 361

Text #2

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.4175. Min Act: -0.1700

Data Index: 1130203 (C4 (Web Text))

Max Activating Token Index: 94

Text #3

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.4332. Min Act: -0.1700

Data Index: 1075371 (C4 (Web Text))

Max Activating Token Index: 627

Text #4

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.3705. Min Act: -0.1700

Data Index: 619653 (C4 (Web Text))

Max Activating Token Index: 21

Text #5

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2920. Min Act: -0.1700

Data Index: 490373 (C4 (Web Text))

Max Activating Token Index: 906

Text #6

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.3234. Min Act: -0.1700

Data Index: 150470 (C4 (Web Text))

Max Activating Token Index: 744

Text #7

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2920. Min Act: -0.1700

Data Index: 1605769 (Python Code)

Max Activating Token Index: 109

Text #8

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2605. Min Act: -0.1700

Data Index: 724753 (C4 (Web Text))

Max Activating Token Index: 243

Text #9

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2448. Min Act: -0.1700

Data Index: 1379761 (Python Code)

Max Activating Token Index: 54

Text #10

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2763. Min Act: -0.1700

Data Index: 1420190 (Python Code)

Max Activating Token Index: 906

Text #11

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2448. Min Act: -0.1700

Data Index: 1650632 (Python Code)

Max Activating Token Index: 281

Text #12

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2605. Min Act: -0.1700

Data Index: 512716 (C4 (Web Text))

Max Activating Token Index: 102

Text #13

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2291. Min Act: -0.1700

Data Index: 485919 (C4 (Web Text))

Max Activating Token Index: 341

Text #14

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2448. Min Act: -0.1700

Data Index: 523980 (C4 (Web Text))

Max Activating Token Index: 364

Text #15

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2448. Min Act: -0.1700

Data Index: 738784 (C4 (Web Text))

Max Activating Token Index: 829

Text #16

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.2134. Min Act: -0.1700

Data Index: 1408337 (Python Code)

Max Activating Token Index: 236

Text #17

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.1976. Min Act: -0.1700

Data Index: 1463779 (Python Code)

Max Activating Token Index: 77

Text #18

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.1976. Min Act: -0.1700

Data Index: 1524470 (Python Code)

Max Activating Token Index: 483

Text #19

Max Range: 3.5742. Min Range: -3.5742

Max Act: 3.1976. Min Act: -0.1700

Data Index: 1102492 (C4 (Web Text))

Max Activating Token Index: 966