Model: GELU Model: 2 Layers, 2048 Neurons per Layer

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

Neuron 2019 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-2l')

Text #0

Max Range: 4.2261. Min Range: -4.2261

Max Act: 4.2261. Min Act: -0.1700

Data Index: 1009514 (C4 (Web Text))

Max Activating Token Index: 802

Text #1

Max Range: 4.2261. Min Range: -4.2261

Max Act: 4.2261. Min Act: -0.1700

Data Index: 633509 (C4 (Web Text))

Max Activating Token Index: 960

Text #2

Max Range: 4.2261. Min Range: -4.2261

Max Act: 4.0698. Min Act: -0.1699

Data Index: 1585840 (Python Code)

Max Activating Token Index: 963

Text #3

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.9291. Min Act: -0.1700

Data Index: 1079728 (C4 (Web Text))

Max Activating Token Index: 558

Text #4

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.9135. Min Act: -0.1700

Data Index: 188712 (C4 (Web Text))

Max Activating Token Index: 124

Text #5

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.9135. Min Act: -0.1700

Data Index: 949760 (C4 (Web Text))

Max Activating Token Index: 185

Text #6

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8822. Min Act: -0.1700

Data Index: 1449067 (Python Code)

Max Activating Token Index: 1023

Text #7

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8353. Min Act: -0.1700

Data Index: 1390244 (Python Code)

Max Activating Token Index: 584

Text #8

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8040. Min Act: -0.1700

Data Index: 1546519 (Python Code)

Max Activating Token Index: 33

Text #9

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8353. Min Act: -0.1693

Data Index: 348461 (C4 (Web Text))

Max Activating Token Index: 1008

Text #10

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8040. Min Act: -0.1700

Data Index: 511554 (C4 (Web Text))

Max Activating Token Index: 571

Text #11

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8197. Min Act: -0.1700

Data Index: 1267815 (C4 (Web Text))

Max Activating Token Index: 310

Text #12

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.7884. Min Act: -0.1700

Data Index: 1579277 (Python Code)

Max Activating Token Index: 1023

Text #13

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8040. Min Act: -0.1700

Data Index: 591100 (C4 (Web Text))

Max Activating Token Index: 244

Text #14

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.7884. Min Act: -0.1700

Data Index: 1574602 (Python Code)

Max Activating Token Index: 21

Text #15

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.7884. Min Act: -0.1700

Data Index: 1408241 (Python Code)

Max Activating Token Index: 20

Text #16

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.7884. Min Act: -0.1700

Data Index: 1430527 (Python Code)

Max Activating Token Index: 20

Text #17

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.7571. Min Act: -0.1700

Data Index: 1424949 (Python Code)

Max Activating Token Index: 414

Text #18

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.8510. Min Act: -0.1700

Data Index: 556811 (C4 (Web Text))

Max Activating Token Index: 262

Text #19

Max Range: 4.2261. Min Range: -4.2261

Max Act: 3.7571. Min Act: -0.1700

Data Index: 674761 (C4 (Web Text))

Max Activating Token Index: 206