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

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

Neuron 260 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.3540. Min Act: -0.1700

Data Index: 1420019 (Python Code)

Max Activating Token Index: 935

Text #1

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.2755. Min Act: -0.1700

Data Index: 1628325 (Python Code)

Max Activating Token Index: 963

Text #2

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.2598. Min Act: -0.1700

Data Index: 1601022 (Python Code)

Max Activating Token Index: 1009

Text #3

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.1654. Min Act: -0.1700

Data Index: 1444969 (Python Code)

Max Activating Token Index: 984

Text #4

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.1969. Min Act: -0.1700

Data Index: 1437892 (Python Code)

Max Activating Token Index: 960

Text #5

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.1497. Min Act: -0.1700

Data Index: 1442014 (Python Code)

Max Activating Token Index: 1000

Text #6

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.1024. Min Act: -0.1700

Data Index: 1461432 (Python Code)

Max Activating Token Index: 946

Text #7

Max Range: 3.3540. Min Range: -3.3540

Max Act: 3.0709. Min Act: -0.1700

Data Index: 743999 (C4 (Web Text))

Max Activating Token Index: 1006

Text #8

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.9761. Min Act: -0.1700

Data Index: 907874 (C4 (Web Text))

Max Activating Token Index: 1015

Text #9

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.9127. Min Act: -0.1700

Data Index: 309609 (C4 (Web Text))

Max Activating Token Index: 977

Text #10

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8810. Min Act: -0.1700

Data Index: 1477032 (Python Code)

Max Activating Token Index: 897

Text #11

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8969. Min Act: -0.1700

Data Index: 1425685 (Python Code)

Max Activating Token Index: 858

Text #12

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8969. Min Act: -0.1700

Data Index: 1334420 (C4 (Web Text))

Max Activating Token Index: 1002

Text #13

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8969. Min Act: -0.1700

Data Index: 1593106 (Python Code)

Max Activating Token Index: 1022

Text #14

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.9127. Min Act: -0.1700

Data Index: 1577530 (Python Code)

Max Activating Token Index: 914

Text #15

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8810. Min Act: -0.1700

Data Index: 193730 (C4 (Web Text))

Max Activating Token Index: 1008

Text #16

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8651. Min Act: -0.1700

Data Index: 950817 (C4 (Web Text))

Max Activating Token Index: 1016

Text #17

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8174. Min Act: -0.1700

Data Index: 278820 (C4 (Web Text))

Max Activating Token Index: 971

Text #18

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8174. Min Act: -0.1700

Data Index: 178244 (C4 (Web Text))

Max Activating Token Index: 1011

Text #19

Max Range: 3.3540. Min Range: -3.3540

Max Act: 2.8333. Min Act: -0.1700

Data Index: 505859 (C4 (Web Text))

Max Activating Token Index: 981