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

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

Neuron 1600 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: 5.6136. Min Range: -5.6136

Max Act: 5.6136. Min Act: -0.1700

Data Index: 498970 (C4 (Web Text))

Max Activating Token Index: 620

Text #1

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.5511. Min Act: -0.1700

Data Index: 1423583 (Python Code)

Max Activating Token Index: 873

Text #2

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.5511. Min Act: -0.1700

Data Index: 1546693 (Python Code)

Max Activating Token Index: 871

Text #3

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.5511. Min Act: -0.1700

Data Index: 838616 (C4 (Web Text))

Max Activating Token Index: 557

Text #4

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.5824. Min Act: -0.1700

Data Index: 1600145 (Python Code)

Max Activating Token Index: 61

Text #5

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.5511. Min Act: -0.1700

Data Index: 1365450 (Python Code)

Max Activating Token Index: 28

Text #6

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.5199. Min Act: -0.1700

Data Index: 1372147 (Python Code)

Max Activating Token Index: 863

Text #7

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4886. Min Act: -0.1700

Data Index: 1318714 (C4 (Web Text))

Max Activating Token Index: 867

Text #8

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4574. Min Act: -0.1700

Data Index: 1417682 (Python Code)

Max Activating Token Index: 29

Text #9

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4574. Min Act: -0.1700

Data Index: 1384658 (Python Code)

Max Activating Token Index: 981

Text #10

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4574. Min Act: -0.1700

Data Index: 1530403 (Python Code)

Max Activating Token Index: 570

Text #11

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4574. Min Act: -0.1700

Data Index: 1638971 (Python Code)

Max Activating Token Index: 872

Text #12

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4886. Min Act: -0.1700

Data Index: 1392504 (Python Code)

Max Activating Token Index: 676

Text #13

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4261. Min Act: -0.1700

Data Index: 1428013 (Python Code)

Max Activating Token Index: 26

Text #14

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.3949. Min Act: -0.1700

Data Index: 1585648 (Python Code)

Max Activating Token Index: 216

Text #15

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.3636. Min Act: -0.1700

Data Index: 354707 (C4 (Web Text))

Max Activating Token Index: 842

Text #16

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.4574. Min Act: -0.1700

Data Index: 307436 (C4 (Web Text))

Max Activating Token Index: 574

Text #17

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.3949. Min Act: -0.1700

Data Index: 1561940 (Python Code)

Max Activating Token Index: 213

Text #18

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.3949. Min Act: -0.1700

Data Index: 1600055 (Python Code)

Max Activating Token Index: 993

Text #19

Max Range: 5.6136. Min Range: -5.6136

Max Act: 5.3636. Min Act: -0.1700

Data Index: 1384984 (Python Code)

Max Activating Token Index: 974