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

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

Neuron 644 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 7.2197. Min Range: -7.2197

Max Act: 7.2197. Min Act: -0.1700

Data Index: 1055714 (C4 (Web Text))

Max Activating Token Index: 994

Text #1

Max Range: 7.2197. Min Range: -7.2197

Max Act: 7.1572. Min Act: -0.1700

Data Index: 1436954 (Python Code)

Max Activating Token Index: 983

Text #2

Max Range: 7.2197. Min Range: -7.2197

Max Act: 7.0009. Min Act: -0.1700

Data Index: 1468395 (Python Code)

Max Activating Token Index: 941

Text #3

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.5634. Min Act: -0.1700

Data Index: 653934 (C4 (Web Text))

Max Activating Token Index: 923

Text #4

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.4384. Min Act: -0.1700

Data Index: 1485842 (Python Code)

Max Activating Token Index: 1014

Text #5

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.4072. Min Act: -0.1700

Data Index: 1564750 (Python Code)

Max Activating Token Index: 986

Text #6

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.4072. Min Act: -0.1699

Data Index: 4874 (C4 (Web Text))

Max Activating Token Index: 882

Text #7

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.3134. Min Act: -0.1700

Data Index: 1130066 (C4 (Web Text))

Max Activating Token Index: 814

Text #8

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.3134. Min Act: -0.1700

Data Index: 1579829 (Python Code)

Max Activating Token Index: 927

Text #9

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.2509. Min Act: -0.1700

Data Index: 491695 (C4 (Web Text))

Max Activating Token Index: 865

Text #10

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.2197. Min Act: -0.1700

Data Index: 913623 (C4 (Web Text))

Max Activating Token Index: 1015

Text #11

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.1572. Min Act: -0.1700

Data Index: 237834 (C4 (Web Text))

Max Activating Token Index: 969

Text #12

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.1572. Min Act: -0.1700

Data Index: 1458836 (Python Code)

Max Activating Token Index: 988

Text #13

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.0947. Min Act: -0.1700

Data Index: 1509406 (Python Code)

Max Activating Token Index: 884

Text #14

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.0947. Min Act: -0.1700

Data Index: 1550606 (Python Code)

Max Activating Token Index: 1007

Text #15

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.0009. Min Act: -0.1700

Data Index: 594673 (C4 (Web Text))

Max Activating Token Index: 967

Text #16

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.0322. Min Act: -0.1700

Data Index: 311104 (C4 (Web Text))

Max Activating Token Index: 1020

Text #17

Max Range: 7.2197. Min Range: -7.2197

Max Act: 6.0009. Min Act: -0.1700

Data Index: 1632553 (Python Code)

Max Activating Token Index: 1022

Text #18

Max Range: 7.2197. Min Range: -7.2197

Max Act: 5.9072. Min Act: -0.1700

Data Index: 1404319 (Python Code)

Max Activating Token Index: 899

Text #19

Max Range: 7.2197. Min Range: -7.2197

Max Act: 5.9072. Min Act: -0.1700

Data Index: 913505 (C4 (Web Text))

Max Activating Token Index: 872