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

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

Neuron 2022 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: 5.4852. Min Range: -5.4852

Max Act: 5.4540. Min Act: -0.1700

Data Index: 606874 (C4 (Web Text))

Max Activating Token Index: 174

Text #1

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.4852. Min Act: -0.1700

Data Index: 1005659 (C4 (Web Text))

Max Activating Token Index: 571

Text #2

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.3290. Min Act: -0.1700

Data Index: 1142851 (C4 (Web Text))

Max Activating Token Index: 555

Text #3

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2977. Min Act: -0.1700

Data Index: 29369 (C4 (Web Text))

Max Activating Token Index: 878

Text #4

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2665. Min Act: -0.1700

Data Index: 439953 (C4 (Web Text))

Max Activating Token Index: 321

Text #5

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.3290. Min Act: -0.1700

Data Index: 919616 (C4 (Web Text))

Max Activating Token Index: 412

Text #6

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.3602. Min Act: -0.1700

Data Index: 629477 (C4 (Web Text))

Max Activating Token Index: 726

Text #7

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2977. Min Act: -0.1700

Data Index: 1102254 (C4 (Web Text))

Max Activating Token Index: 907

Text #8

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2665. Min Act: -0.1700

Data Index: 981149 (C4 (Web Text))

Max Activating Token Index: 176

Text #9

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2352. Min Act: -0.1700

Data Index: 442318 (C4 (Web Text))

Max Activating Token Index: 552

Text #10

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2665. Min Act: -0.1700

Data Index: 485289 (C4 (Web Text))

Max Activating Token Index: 810

Text #11

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2040. Min Act: -0.1700

Data Index: 546504 (C4 (Web Text))

Max Activating Token Index: 211

Text #12

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2040. Min Act: -0.1700

Data Index: 808601 (C4 (Web Text))

Max Activating Token Index: 259

Text #13

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2665. Min Act: -0.1700

Data Index: 253295 (C4 (Web Text))

Max Activating Token Index: 990

Text #14

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2977. Min Act: -0.1700

Data Index: 833605 (C4 (Web Text))

Max Activating Token Index: 860

Text #15

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.1727. Min Act: -0.1700

Data Index: 564361 (C4 (Web Text))

Max Activating Token Index: 847

Text #16

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2352. Min Act: -0.1700

Data Index: 573684 (C4 (Web Text))

Max Activating Token Index: 387

Text #17

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2040. Min Act: -0.1700

Data Index: 247146 (C4 (Web Text))

Max Activating Token Index: 681

Text #18

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.2040. Min Act: -0.1700

Data Index: 295574 (C4 (Web Text))

Max Activating Token Index: 682

Text #19

Max Range: 5.4852. Min Range: -5.4852

Max Act: 5.1727. Min Act: -0.1700

Data Index: 19483 (C4 (Web Text))

Max Activating Token Index: 253