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

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

Neuron 2027 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: 3.8855. Min Range: -3.8855

Max Act: 3.8855. Min Act: -0.1700

Data Index: 692361 (C4 (Web Text))

Max Activating Token Index: 604

Text #1

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5726. Min Act: -0.1700

Data Index: 623654 (C4 (Web Text))

Max Activating Token Index: 703

Text #2

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.6665. Min Act: -0.1700

Data Index: 1161720 (C4 (Web Text))

Max Activating Token Index: 627

Text #3

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5100. Min Act: -0.1700

Data Index: 1229204 (C4 (Web Text))

Max Activating Token Index: 516

Text #4

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5726. Min Act: -0.1700

Data Index: 785500 (C4 (Web Text))

Max Activating Token Index: 491

Text #5

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.4786. Min Act: -0.1700

Data Index: 223480 (C4 (Web Text))

Max Activating Token Index: 458

Text #6

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5726. Min Act: -0.1700

Data Index: 50517 (C4 (Web Text))

Max Activating Token Index: 652

Text #7

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5726. Min Act: -0.1700

Data Index: 705228 (C4 (Web Text))

Max Activating Token Index: 271

Text #8

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.4786. Min Act: -0.1700

Data Index: 864076 (C4 (Web Text))

Max Activating Token Index: 790

Text #9

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.6039. Min Act: -0.1700

Data Index: 696346 (C4 (Web Text))

Max Activating Token Index: 435

Text #10

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5100. Min Act: -0.1700

Data Index: 737192 (C4 (Web Text))

Max Activating Token Index: 287

Text #11

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.3845. Min Act: -0.1700

Data Index: 961888 (C4 (Web Text))

Max Activating Token Index: 614

Text #12

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.4786. Min Act: -0.1700

Data Index: 25381 (C4 (Web Text))

Max Activating Token Index: 901

Text #13

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.4786. Min Act: -0.1700

Data Index: 201305 (C4 (Web Text))

Max Activating Token Index: 810

Text #14

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.4473. Min Act: -0.1700

Data Index: 933067 (C4 (Web Text))

Max Activating Token Index: 551

Text #15

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5100. Min Act: -0.1700

Data Index: 363103 (C4 (Web Text))

Max Activating Token Index: 932

Text #16

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5100. Min Act: -0.1700

Data Index: 1478861 (Python Code)

Max Activating Token Index: 578

Text #17

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5100. Min Act: -0.1700

Data Index: 532459 (C4 (Web Text))

Max Activating Token Index: 778

Text #18

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.5100. Min Act: -0.1700

Data Index: 34494 (C4 (Web Text))

Max Activating Token Index: 879

Text #19

Max Range: 3.8855. Min Range: -3.8855

Max Act: 3.4473. Min Act: -0.1700

Data Index: 834516 (C4 (Web Text))

Max Activating Token Index: 157