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

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

Neuron 558 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: 5.4955. Min Range: -5.4955

Max Act: 5.4955. Min Act: -0.1700

Data Index: 333880 (C4 (Web Text))

Max Activating Token Index: 930

Text #1

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.4017. Min Act: -0.1700

Data Index: 66255 (C4 (Web Text))

Max Activating Token Index: 994

Text #2

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.4330. Min Act: -0.1700

Data Index: 179267 (C4 (Web Text))

Max Activating Token Index: 575

Text #3

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.4017. Min Act: -0.1700

Data Index: 156974 (C4 (Web Text))

Max Activating Token Index: 287

Text #4

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.4330. Min Act: -0.1700

Data Index: 213850 (C4 (Web Text))

Max Activating Token Index: 237

Text #5

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3705. Min Act: -0.1700

Data Index: 594087 (C4 (Web Text))

Max Activating Token Index: 444

Text #6

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3705. Min Act: -0.1700

Data Index: 1267324 (C4 (Web Text))

Max Activating Token Index: 966

Text #7

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3705. Min Act: -0.1700

Data Index: 211805 (C4 (Web Text))

Max Activating Token Index: 472

Text #8

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3392. Min Act: -0.1700

Data Index: 127091 (C4 (Web Text))

Max Activating Token Index: 232

Text #9

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3392. Min Act: -0.1700

Data Index: 1068405 (C4 (Web Text))

Max Activating Token Index: 124

Text #10

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3392. Min Act: -0.1700

Data Index: 772791 (C4 (Web Text))

Max Activating Token Index: 522

Text #11

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.2767. Min Act: -0.1700

Data Index: 260133 (C4 (Web Text))

Max Activating Token Index: 542

Text #12

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.2767. Min Act: -0.1700

Data Index: 247743 (C4 (Web Text))

Max Activating Token Index: 213

Text #13

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.2767. Min Act: -0.1699

Data Index: 343802 (C4 (Web Text))

Max Activating Token Index: 800

Text #14

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3392. Min Act: -0.1700

Data Index: 666065 (C4 (Web Text))

Max Activating Token Index: 546

Text #15

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.2767. Min Act: -0.1700

Data Index: 171316 (C4 (Web Text))

Max Activating Token Index: 970

Text #16

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.3080. Min Act: -0.1700

Data Index: 553966 (C4 (Web Text))

Max Activating Token Index: 1017

Text #17

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.2767. Min Act: -0.1700

Data Index: 184301 (C4 (Web Text))

Max Activating Token Index: 142

Text #18

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.2767. Min Act: -0.1699

Data Index: 1638554 (Python Code)

Max Activating Token Index: 453

Text #19

Max Range: 5.4955. Min Range: -5.4955

Max Act: 5.2455. Min Act: -0.1699

Data Index: 889807 (C4 (Web Text))

Max Activating Token Index: 865