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

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

Neuron 92 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: 3.4463. Min Range: -3.4463

Max Act: 3.4463. Min Act: -0.1700

Data Index: 115338 (C4 (Web Text))

Max Activating Token Index: 433

Text #1

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.3836. Min Act: -0.1700

Data Index: 1003737 (C4 (Web Text))

Max Activating Token Index: 188

Text #2

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.3836. Min Act: -0.1700

Data Index: 134102 (C4 (Web Text))

Max Activating Token Index: 571

Text #3

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.3522. Min Act: -0.1700

Data Index: 534680 (C4 (Web Text))

Max Activating Token Index: 144

Text #4

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.3208. Min Act: -0.1700

Data Index: 210244 (C4 (Web Text))

Max Activating Token Index: 177

Text #5

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2894. Min Act: -0.1700

Data Index: 361619 (C4 (Web Text))

Max Activating Token Index: 536

Text #6

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2894. Min Act: -0.1700

Data Index: 1605330 (Python Code)

Max Activating Token Index: 339

Text #7

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2894. Min Act: -0.1700

Data Index: 729154 (C4 (Web Text))

Max Activating Token Index: 857

Text #8

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2266. Min Act: -0.1700

Data Index: 964848 (C4 (Web Text))

Max Activating Token Index: 559

Text #9

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2894. Min Act: -0.1700

Data Index: 743993 (C4 (Web Text))

Max Activating Token Index: 858

Text #10

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2266. Min Act: -0.1700

Data Index: 914416 (C4 (Web Text))

Max Activating Token Index: 229

Text #11

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2423. Min Act: -0.1700

Data Index: 865918 (C4 (Web Text))

Max Activating Token Index: 705

Text #12

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2108. Min Act: -0.1700

Data Index: 589158 (C4 (Web Text))

Max Activating Token Index: 598

Text #13

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2580. Min Act: -0.1700

Data Index: 1416878 (Python Code)

Max Activating Token Index: 171

Text #14

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2108. Min Act: -0.1700

Data Index: 919165 (C4 (Web Text))

Max Activating Token Index: 176

Text #15

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2266. Min Act: -0.1700

Data Index: 1178520 (C4 (Web Text))

Max Activating Token Index: 138

Text #16

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.1794. Min Act: -0.1700

Data Index: 1345581 (C4 (Web Text))

Max Activating Token Index: 42

Text #17

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.1794. Min Act: -0.1700

Data Index: 689425 (C4 (Web Text))

Max Activating Token Index: 158

Text #18

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.2108. Min Act: -0.1700

Data Index: 1487267 (Python Code)

Max Activating Token Index: 365

Text #19

Max Range: 3.4463. Min Range: -3.4463

Max Act: 3.1951. Min Act: -0.1700

Data Index: 584632 (C4 (Web Text))

Max Activating Token Index: 99