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

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

Neuron 711 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: 5.4755. Min Range: -5.4755

Max Act: 5.4755. Min Act: -0.1700

Data Index: 1589308 (Python Code)

Max Activating Token Index: 589

Text #1

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.1943. Min Act: -0.1699

Data Index: 489552 (C4 (Web Text))

Max Activating Token Index: 698

Text #2

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.1318. Min Act: -0.1700

Data Index: 674063 (C4 (Web Text))

Max Activating Token Index: 409

Text #3

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.1005. Min Act: -0.1700

Data Index: 1432535 (Python Code)

Max Activating Token Index: 480

Text #4

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.0380. Min Act: -0.1699

Data Index: 1467333 (Python Code)

Max Activating Token Index: 363

Text #5

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.1318. Min Act: -0.1700

Data Index: 467844 (C4 (Web Text))

Max Activating Token Index: 234

Text #6

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.1005. Min Act: -0.1700

Data Index: 598990 (C4 (Web Text))

Max Activating Token Index: 430

Text #7

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.0693. Min Act: -0.1698

Data Index: 879333 (C4 (Web Text))

Max Activating Token Index: 414

Text #8

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.0068. Min Act: -0.1699

Data Index: 268062 (C4 (Web Text))

Max Activating Token Index: 975

Text #9

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.0068. Min Act: -0.1700

Data Index: 1588481 (Python Code)

Max Activating Token Index: 577

Text #10

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9755. Min Act: -0.1700

Data Index: 1503750 (Python Code)

Max Activating Token Index: 808

Text #11

Max Range: 5.4755. Min Range: -5.4755

Max Act: 5.0068. Min Act: -0.1699

Data Index: 373810 (C4 (Web Text))

Max Activating Token Index: 794

Text #12

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9755. Min Act: -0.1700

Data Index: 206529 (C4 (Web Text))

Max Activating Token Index: 859

Text #13

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9130. Min Act: -0.1700

Data Index: 1204620 (C4 (Web Text))

Max Activating Token Index: 665

Text #14

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9755. Min Act: -0.1700

Data Index: 1463823 (Python Code)

Max Activating Token Index: 538

Text #15

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9130. Min Act: -0.1700

Data Index: 1619813 (Python Code)

Max Activating Token Index: 332

Text #16

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9755. Min Act: -0.1700

Data Index: 1497345 (Python Code)

Max Activating Token Index: 860

Text #17

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9130. Min Act: -0.1700

Data Index: 206249 (C4 (Web Text))

Max Activating Token Index: 830

Text #18

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.8818. Min Act: -0.1700

Data Index: 402238 (C4 (Web Text))

Max Activating Token Index: 930

Text #19

Max Range: 5.4755. Min Range: -5.4755

Max Act: 4.9130. Min Act: -0.1700

Data Index: 1627680 (Python Code)

Max Activating Token Index: 990