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

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

Neuron 1118 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.4954. Min Act: -0.1700

Data Index: 1357686 (C4 (Web Text))

Max Activating Token Index: 975

Text #1

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.4016. Min Act: -0.1700

Data Index: 1447913 (Python Code)

Max Activating Token Index: 350

Text #2

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.3079. Min Act: -0.1700

Data Index: 1496838 (Python Code)

Max Activating Token Index: 956

Text #3

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.3391. Min Act: -0.1700

Data Index: 1455787 (Python Code)

Max Activating Token Index: 1008

Text #4

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.3704. Min Act: -0.1700

Data Index: 1507285 (Python Code)

Max Activating Token Index: 388

Text #5

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1204. Min Act: -0.1700

Data Index: 415629 (C4 (Web Text))

Max Activating Token Index: 879

Text #6

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.3391. Min Act: -0.1700

Data Index: 1584535 (Python Code)

Max Activating Token Index: 544

Text #7

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.2766. Min Act: -0.1700

Data Index: 1363603 (Python Code)

Max Activating Token Index: 1010

Text #8

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.2766. Min Act: -0.1700

Data Index: 726793 (C4 (Web Text))

Max Activating Token Index: 846

Text #9

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.2454. Min Act: -0.1700

Data Index: 1451536 (Python Code)

Max Activating Token Index: 267

Text #10

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.2454. Min Act: -0.1700

Data Index: 1327355 (C4 (Web Text))

Max Activating Token Index: 868

Text #11

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1829. Min Act: -0.1700

Data Index: 1329104 (C4 (Web Text))

Max Activating Token Index: 792

Text #12

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.2454. Min Act: -0.1700

Data Index: 1648911 (Python Code)

Max Activating Token Index: 847

Text #13

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.2141. Min Act: -0.1700

Data Index: 1522437 (Python Code)

Max Activating Token Index: 967

Text #14

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1516. Min Act: -0.1700

Data Index: 448473 (C4 (Web Text))

Max Activating Token Index: 484

Text #15

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1829. Min Act: -0.1700

Data Index: 1461221 (Python Code)

Max Activating Token Index: 281

Text #16

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1829. Min Act: -0.1700

Data Index: 1194759 (C4 (Web Text))

Max Activating Token Index: 889

Text #17

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1516. Min Act: -0.1700

Data Index: 1569305 (Python Code)

Max Activating Token Index: 546

Text #18

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1829. Min Act: -0.1700

Data Index: 1476080 (Python Code)

Max Activating Token Index: 523

Text #19

Max Range: 5.4954. Min Range: -5.4954

Max Act: 5.1516. Min Act: -0.1700

Data Index: 1567207 (Python Code)

Max Activating Token Index: 245