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

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

Neuron 1573 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.9691. Min Act: -0.1700

Data Index: 113850 (C4 (Web Text))

Max Activating Token Index: 532

Text #1

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.8440. Min Act: -0.1700

Data Index: 1236468 (C4 (Web Text))

Max Activating Token Index: 359

Text #2

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.6876. Min Act: -0.1700

Data Index: 1063179 (C4 (Web Text))

Max Activating Token Index: 446

Text #3

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7188. Min Act: -0.1700

Data Index: 1233746 (C4 (Web Text))

Max Activating Token Index: 159

Text #4

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7501. Min Act: -0.1700

Data Index: 635337 (C4 (Web Text))

Max Activating Token Index: 119

Text #5

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7658. Min Act: -0.1700

Data Index: 407695 (C4 (Web Text))

Max Activating Token Index: 441

Text #6

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7814. Min Act: -0.1700

Data Index: 534263 (C4 (Web Text))

Max Activating Token Index: 558

Text #7

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7188. Min Act: -0.1700

Data Index: 310028 (C4 (Web Text))

Max Activating Token Index: 449

Text #8

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7501. Min Act: -0.1700

Data Index: 395846 (C4 (Web Text))

Max Activating Token Index: 655

Text #9

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7188. Min Act: -0.1700

Data Index: 303923 (C4 (Web Text))

Max Activating Token Index: 900

Text #10

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7188. Min Act: -0.1700

Data Index: 1116174 (C4 (Web Text))

Max Activating Token Index: 986

Text #11

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7188. Min Act: -0.1700

Data Index: 1112499 (C4 (Web Text))

Max Activating Token Index: 506

Text #12

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.6563. Min Act: -0.1700

Data Index: 60328 (C4 (Web Text))

Max Activating Token Index: 506

Text #13

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7032. Min Act: -0.1700

Data Index: 696491 (C4 (Web Text))

Max Activating Token Index: 322

Text #14

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7188. Min Act: -0.1700

Data Index: 512921 (C4 (Web Text))

Max Activating Token Index: 456

Text #15

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.7188. Min Act: -0.1700

Data Index: 506666 (C4 (Web Text))

Max Activating Token Index: 583

Text #16

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.6563. Min Act: -0.1700

Data Index: 22149 (C4 (Web Text))

Max Activating Token Index: 490

Text #17

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.6719. Min Act: -0.1700

Data Index: 1560737 (Python Code)

Max Activating Token Index: 944

Text #18

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.6093. Min Act: -0.1700

Data Index: 309004 (C4 (Web Text))

Max Activating Token Index: 607

Text #19

Max Range: 3.9691. Min Range: -3.9691

Max Act: 3.6876. Min Act: -0.1700

Data Index: 79786 (C4 (Web Text))

Max Activating Token Index: 208