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

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

Neuron 58 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: 5.4403. Min Range: -5.4403

Max Act: 5.4403. Min Act: -0.1700

Data Index: 931492 (C4 (Web Text))

Max Activating Token Index: 579

Text #1

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.2059. Min Act: -0.1700

Data Index: 1229637 (C4 (Web Text))

Max Activating Token Index: 118

Text #2

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.2059. Min Act: -0.1700

Data Index: 1154382 (C4 (Web Text))

Max Activating Token Index: 255

Text #3

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.1434. Min Act: -0.1700

Data Index: 193028 (C4 (Web Text))

Max Activating Token Index: 287

Text #4

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.0653. Min Act: -0.1700

Data Index: 1328887 (C4 (Web Text))

Max Activating Token Index: 928

Text #5

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.0496. Min Act: -0.1700

Data Index: 723474 (C4 (Web Text))

Max Activating Token Index: 935

Text #6

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.0340. Min Act: -0.1700

Data Index: 317218 (C4 (Web Text))

Max Activating Token Index: 296

Text #7

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.0340. Min Act: -0.1699

Data Index: 690809 (C4 (Web Text))

Max Activating Token Index: 616

Text #8

Max Range: 5.4403. Min Range: -5.4403

Max Act: 5.0028. Min Act: -0.1699

Data Index: 845996 (C4 (Web Text))

Max Activating Token Index: 118

Text #9

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9403. Min Act: -0.1699

Data Index: 939753 (C4 (Web Text))

Max Activating Token Index: 676

Text #10

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9715. Min Act: -0.1699

Data Index: 136774 (C4 (Web Text))

Max Activating Token Index: 233

Text #11

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9871. Min Act: -0.1700

Data Index: 935352 (C4 (Web Text))

Max Activating Token Index: 582

Text #12

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9246. Min Act: -0.1700

Data Index: 1500364 (Python Code)

Max Activating Token Index: 91

Text #13

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9403. Min Act: -0.1700

Data Index: 1116503 (C4 (Web Text))

Max Activating Token Index: 244

Text #14

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9246. Min Act: -0.1700

Data Index: 256135 (C4 (Web Text))

Max Activating Token Index: 414

Text #15

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9403. Min Act: -0.1700

Data Index: 542363 (C4 (Web Text))

Max Activating Token Index: 708

Text #16

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.8934. Min Act: -0.1694

Data Index: 252936 (C4 (Web Text))

Max Activating Token Index: 284

Text #17

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.9715. Min Act: -0.1700

Data Index: 426173 (C4 (Web Text))

Max Activating Token Index: 512

Text #18

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.8934. Min Act: -0.1700

Data Index: 636161 (C4 (Web Text))

Max Activating Token Index: 297

Text #19

Max Range: 5.4403. Min Range: -5.4403

Max Act: 4.8934. Min Act: -0.1700

Data Index: 1119985 (C4 (Web Text))

Max Activating Token Index: 545