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

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

Neuron 201 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: 5.6877. Min Range: -5.6877

Max Act: 5.6877. Min Act: -0.1700

Data Index: 388354 (C4 (Web Text))

Max Activating Token Index: 927

Text #1

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.4689. Min Act: -0.1700

Data Index: 1015254 (C4 (Web Text))

Max Activating Token Index: 849

Text #2

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.4064. Min Act: -0.1700

Data Index: 936507 (C4 (Web Text))

Max Activating Token Index: 678

Text #3

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.2814. Min Act: -0.1700

Data Index: 1032709 (C4 (Web Text))

Max Activating Token Index: 928

Text #4

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.3439. Min Act: -0.1700

Data Index: 439217 (C4 (Web Text))

Max Activating Token Index: 720

Text #5

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.1877. Min Act: -0.1700

Data Index: 153130 (C4 (Web Text))

Max Activating Token Index: 726

Text #6

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.3127. Min Act: -0.1700

Data Index: 861989 (C4 (Web Text))

Max Activating Token Index: 462

Text #7

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.2502. Min Act: -0.1700

Data Index: 948907 (C4 (Web Text))

Max Activating Token Index: 520

Text #8

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.2189. Min Act: -0.1700

Data Index: 194820 (C4 (Web Text))

Max Activating Token Index: 225

Text #9

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.1564. Min Act: -0.1700

Data Index: 1204687 (C4 (Web Text))

Max Activating Token Index: 948

Text #10

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.1564. Min Act: -0.1700

Data Index: 1356001 (C4 (Web Text))

Max Activating Token Index: 669

Text #11

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.1564. Min Act: -0.1700

Data Index: 1206975 (C4 (Web Text))

Max Activating Token Index: 927

Text #12

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.0939. Min Act: -0.1700

Data Index: 1550139 (Python Code)

Max Activating Token Index: 756

Text #13

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.0002. Min Act: -0.1700

Data Index: 922482 (C4 (Web Text))

Max Activating Token Index: 796

Text #14

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.0314. Min Act: -0.1700

Data Index: 566086 (C4 (Web Text))

Max Activating Token Index: 589

Text #15

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.0314. Min Act: -0.1700

Data Index: 74251 (C4 (Web Text))

Max Activating Token Index: 366

Text #16

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.0314. Min Act: -0.1700

Data Index: 1006097 (C4 (Web Text))

Max Activating Token Index: 436

Text #17

Max Range: 5.6877. Min Range: -5.6877

Max Act: 4.9689. Min Act: -0.1700

Data Index: 1354232 (C4 (Web Text))

Max Activating Token Index: 573

Text #18

Max Range: 5.6877. Min Range: -5.6877

Max Act: 5.0002. Min Act: -0.1700

Data Index: 1349572 (C4 (Web Text))

Max Activating Token Index: 279

Text #19

Max Range: 5.6877. Min Range: -5.6877

Max Act: 4.9377. Min Act: -0.1700

Data Index: 1092059 (C4 (Web Text))

Max Activating Token Index: 869