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

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

Neuron 1887 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: 3.9112. Min Range: -3.9112

Max Act: 3.9112. Min Act: -0.1700

Data Index: 1331515 (C4 (Web Text))

Max Activating Token Index: 128

Text #1

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.8487. Min Act: -0.1700

Data Index: 574533 (C4 (Web Text))

Max Activating Token Index: 132

Text #2

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.8330. Min Act: -0.1700

Data Index: 255106 (C4 (Web Text))

Max Activating Token Index: 174

Text #3

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.8018. Min Act: -0.1700

Data Index: 139179 (C4 (Web Text))

Max Activating Token Index: 347

Text #4

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7392. Min Act: -0.1700

Data Index: 1366757 (Python Code)

Max Activating Token Index: 929

Text #5

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7548. Min Act: -0.1700

Data Index: 308267 (C4 (Web Text))

Max Activating Token Index: 231

Text #6

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7705. Min Act: -0.1700

Data Index: 145874 (C4 (Web Text))

Max Activating Token Index: 160

Text #7

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7548. Min Act: -0.1700

Data Index: 297044 (C4 (Web Text))

Max Activating Token Index: 942

Text #8

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7705. Min Act: -0.1700

Data Index: 891325 (C4 (Web Text))

Max Activating Token Index: 982

Text #9

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7548. Min Act: -0.1700

Data Index: 1436442 (Python Code)

Max Activating Token Index: 745

Text #10

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7079. Min Act: -0.1700

Data Index: 1622080 (Python Code)

Max Activating Token Index: 689

Text #11

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7705. Min Act: -0.1700

Data Index: 1547003 (Python Code)

Max Activating Token Index: 545

Text #12

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7548. Min Act: -0.1700

Data Index: 812668 (C4 (Web Text))

Max Activating Token Index: 1023

Text #13

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7705. Min Act: -0.1700

Data Index: 976786 (C4 (Web Text))

Max Activating Token Index: 144

Text #14

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7235. Min Act: -0.1700

Data Index: 1392819 (Python Code)

Max Activating Token Index: 1023

Text #15

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.6922. Min Act: -0.1700

Data Index: 1259539 (C4 (Web Text))

Max Activating Token Index: 342

Text #16

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.6922. Min Act: -0.1700

Data Index: 130538 (C4 (Web Text))

Max Activating Token Index: 177

Text #17

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7235. Min Act: -0.1700

Data Index: 213530 (C4 (Web Text))

Max Activating Token Index: 276

Text #18

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.6296. Min Act: -0.1700

Data Index: 750139 (C4 (Web Text))

Max Activating Token Index: 933

Text #19

Max Range: 3.9112. Min Range: -3.9112

Max Act: 3.7235. Min Act: -0.1700

Data Index: 680472 (C4 (Web Text))

Max Activating Token Index: 337