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

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

Neuron 978 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.4830. Min Act: -0.1700

Data Index: 157277 (C4 (Web Text))

Max Activating Token Index: 131

Text #1

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.4830. Min Act: -0.1700

Data Index: 1404360 (Python Code)

Max Activating Token Index: 763

Text #2

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.4046. Min Act: -0.1700

Data Index: 269333 (C4 (Web Text))

Max Activating Token Index: 606

Text #3

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.4360. Min Act: -0.1700

Data Index: 1474918 (Python Code)

Max Activating Token Index: 336

Text #4

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.4360. Min Act: -0.1700

Data Index: 1619999 (Python Code)

Max Activating Token Index: 917

Text #5

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3262. Min Act: -0.1700

Data Index: 1414677 (Python Code)

Max Activating Token Index: 436

Text #6

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3889. Min Act: -0.1700

Data Index: 1539732 (Python Code)

Max Activating Token Index: 581

Text #7

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.4046. Min Act: -0.1700

Data Index: 1271894 (C4 (Web Text))

Max Activating Token Index: 592

Text #8

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3576. Min Act: -0.1700

Data Index: 1645679 (Python Code)

Max Activating Token Index: 667

Text #9

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3733. Min Act: -0.1700

Data Index: 1081734 (C4 (Web Text))

Max Activating Token Index: 759

Text #10

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3419. Min Act: -0.1700

Data Index: 1495654 (Python Code)

Max Activating Token Index: 915

Text #11

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3576. Min Act: -0.1700

Data Index: 1567866 (Python Code)

Max Activating Token Index: 956

Text #12

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3105. Min Act: -0.1700

Data Index: 1010247 (C4 (Web Text))

Max Activating Token Index: 822

Text #13

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3262. Min Act: -0.1700

Data Index: 1610671 (Python Code)

Max Activating Token Index: 107

Text #14

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.2948. Min Act: -0.1700

Data Index: 1534971 (Python Code)

Max Activating Token Index: 426

Text #15

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3576. Min Act: -0.1700

Data Index: 1656887 (Python Code)

Max Activating Token Index: 788

Text #16

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.2791. Min Act: -0.1700

Data Index: 1545632 (Python Code)

Max Activating Token Index: 258

Text #17

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3262. Min Act: -0.1700

Data Index: 401784 (C4 (Web Text))

Max Activating Token Index: 373

Text #18

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.3105. Min Act: -0.1700

Data Index: 1255679 (C4 (Web Text))

Max Activating Token Index: 255

Text #19

Max Range: 3.4830. Min Range: -3.4830

Max Act: 3.2948. Min Act: -0.1700

Data Index: 217811 (C4 (Web Text))

Max Activating Token Index: 234