Model: GELU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 480 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.4188. Min Act: -0.1700

Data Index: 1090179 (C4 (Web Text))

Max Activating Token Index: 1022

Text #1

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.4658. Min Act: -0.1700

Data Index: 666289 (C4 (Web Text))

Max Activating Token Index: 654

Text #2

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.3717. Min Act: -0.1700

Data Index: 241431 (C4 (Web Text))

Max Activating Token Index: 393

Text #3

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.3403. Min Act: -0.1700

Data Index: 969802 (C4 (Web Text))

Max Activating Token Index: 524

Text #4

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.3403. Min Act: -0.1700

Data Index: 597006 (C4 (Web Text))

Max Activating Token Index: 470

Text #5

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.3089. Min Act: -0.1700

Data Index: 539572 (C4 (Web Text))

Max Activating Token Index: 646

Text #6

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.3089. Min Act: -0.1700

Data Index: 1285092 (C4 (Web Text))

Max Activating Token Index: 109

Text #7

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.3089. Min Act: -0.1700

Data Index: 1107771 (C4 (Web Text))

Max Activating Token Index: 818

Text #8

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.2461. Min Act: -0.1700

Data Index: 220733 (C4 (Web Text))

Max Activating Token Index: 840

Text #9

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.2932. Min Act: -0.1700

Data Index: 1193797 (C4 (Web Text))

Max Activating Token Index: 790

Text #10

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.2932. Min Act: -0.1700

Data Index: 1065850 (C4 (Web Text))

Max Activating Token Index: 56

Text #11

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.2146. Min Act: -0.1700

Data Index: 1112487 (C4 (Web Text))

Max Activating Token Index: 831

Text #12

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.2618. Min Act: -0.1700

Data Index: 1273254 (C4 (Web Text))

Max Activating Token Index: 701

Text #13

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.1517. Min Act: -0.1700

Data Index: 977220 (C4 (Web Text))

Max Activating Token Index: 788

Text #14

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.2304. Min Act: -0.1700

Data Index: 733035 (C4 (Web Text))

Max Activating Token Index: 959

Text #15

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.1674. Min Act: -0.1700

Data Index: 760135 (C4 (Web Text))

Max Activating Token Index: 207

Text #16

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.2146. Min Act: -0.1700

Data Index: 1439195 (Python Code)

Max Activating Token Index: 590

Text #17

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.1674. Min Act: -0.1700

Data Index: 1062146 (C4 (Web Text))

Max Activating Token Index: 314

Text #18

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.1674. Min Act: -0.1700

Data Index: 261381 (C4 (Web Text))

Max Activating Token Index: 1014

Text #19

Max Range: 3.4658. Min Range: -3.4658

Max Act: 3.0886. Min Act: -0.1700

Data Index: 148627 (C4 (Web Text))

Max Activating Token Index: 726