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

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

Neuron 1596 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.3810. Min Act: -0.1700

Data Index: 957609 (C4 (Web Text))

Max Activating Token Index: 271

Text #1

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.2247. Min Act: -0.1700

Data Index: 367948 (C4 (Web Text))

Max Activating Token Index: 50

Text #2

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.2247. Min Act: -0.1700

Data Index: 1077806 (C4 (Web Text))

Max Activating Token Index: 846

Text #3

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0997. Min Act: -0.1700

Data Index: 1285895 (C4 (Web Text))

Max Activating Token Index: 411

Text #4

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.1622. Min Act: -0.1700

Data Index: 349216 (C4 (Web Text))

Max Activating Token Index: 257

Text #5

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.1310. Min Act: -0.1700

Data Index: 613055 (C4 (Web Text))

Max Activating Token Index: 623

Text #6

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.1310. Min Act: -0.1055

Data Index: 529099 (C4 (Web Text))

Max Activating Token Index: 676

Text #7

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0685. Min Act: -0.1700

Data Index: 630682 (C4 (Web Text))

Max Activating Token Index: 501

Text #8

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0372. Min Act: -0.1700

Data Index: 796791 (C4 (Web Text))

Max Activating Token Index: 754

Text #9

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.1310. Min Act: -0.1700

Data Index: 307984 (C4 (Web Text))

Max Activating Token Index: 124

Text #10

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.1310. Min Act: -0.1700

Data Index: 355922 (C4 (Web Text))

Max Activating Token Index: 489

Text #11

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0997. Min Act: -0.1700

Data Index: 222860 (C4 (Web Text))

Max Activating Token Index: 361

Text #12

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0685. Min Act: -0.1700

Data Index: 1400341 (Python Code)

Max Activating Token Index: 27

Text #13

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0997. Min Act: -0.1700

Data Index: 629863 (C4 (Web Text))

Max Activating Token Index: 442

Text #14

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0685. Min Act: -0.1700

Data Index: 1122491 (C4 (Web Text))

Max Activating Token Index: 431

Text #15

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0997. Min Act: -0.1700

Data Index: 308064 (C4 (Web Text))

Max Activating Token Index: 960

Text #16

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0372. Min Act: -0.1700

Data Index: 1329533 (C4 (Web Text))

Max Activating Token Index: 863

Text #17

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0372. Min Act: -0.1700

Data Index: 125321 (C4 (Web Text))

Max Activating Token Index: 143

Text #18

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0372. Min Act: -0.1700

Data Index: 593835 (C4 (Web Text))

Max Activating Token Index: 296

Text #19

Max Range: 5.3810. Min Range: -5.3810

Max Act: 5.0060. Min Act: -0.1700

Data Index: 385733 (C4 (Web Text))

Max Activating Token Index: 641