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

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

Neuron 897 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.6979. Min Act: -0.1700

Data Index: 1519312 (Python Code)

Max Activating Token Index: 1023

Text #1

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5727. Min Act: -0.1700

Data Index: 1311867 (C4 (Web Text))

Max Activating Token Index: 805

Text #2

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.6040. Min Act: -0.1700

Data Index: 1408829 (Python Code)

Max Activating Token Index: 165

Text #3

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5414. Min Act: -0.1700

Data Index: 1422195 (Python Code)

Max Activating Token Index: 1023

Text #4

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5414. Min Act: -0.1700

Data Index: 1656266 (Python Code)

Max Activating Token Index: 200

Text #5

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5101. Min Act: -0.1700

Data Index: 1518896 (Python Code)

Max Activating Token Index: 126

Text #6

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5101. Min Act: -0.1700

Data Index: 531144 (C4 (Web Text))

Max Activating Token Index: 564

Text #7

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5101. Min Act: -0.1700

Data Index: 1534205 (Python Code)

Max Activating Token Index: 116

Text #8

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5101. Min Act: -0.1700

Data Index: 265140 (C4 (Web Text))

Max Activating Token Index: 906

Text #9

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4787. Min Act: -0.1700

Data Index: 1429076 (Python Code)

Max Activating Token Index: 890

Text #10

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4787. Min Act: -0.1700

Data Index: 1505116 (Python Code)

Max Activating Token Index: 1023

Text #11

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4787. Min Act: -0.1700

Data Index: 1398897 (Python Code)

Max Activating Token Index: 1023

Text #12

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4630. Min Act: -0.1700

Data Index: 1587127 (Python Code)

Max Activating Token Index: 1023

Text #13

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4630. Min Act: -0.1700

Data Index: 416744 (C4 (Web Text))

Max Activating Token Index: 488

Text #14

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4630. Min Act: -0.1700

Data Index: 1296123 (C4 (Web Text))

Max Activating Token Index: 865

Text #15

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4474. Min Act: -0.1700

Data Index: 546176 (C4 (Web Text))

Max Activating Token Index: 81

Text #16

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.5101. Min Act: -0.1700

Data Index: 1492504 (Python Code)

Max Activating Token Index: 1023

Text #17

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4787. Min Act: -0.1700

Data Index: 1616788 (Python Code)

Max Activating Token Index: 1023

Text #18

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4787. Min Act: -0.1700

Data Index: 1593488 (Python Code)

Max Activating Token Index: 687

Text #19

Max Range: 3.6979. Min Range: -3.6979

Max Act: 3.4630. Min Act: -0.1700

Data Index: 527251 (C4 (Web Text))

Max Activating Token Index: 716