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

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

Neuron 238 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4546. Min Act: -0.1700

Data Index: 1121176 (C4 (Web Text))

Max Activating Token Index: 995

Text #1

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4859. Min Act: -0.1700

Data Index: 1372555 (Python Code)

Max Activating Token Index: 555

Text #2

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4075. Min Act: -0.1700

Data Index: 1473240 (Python Code)

Max Activating Token Index: 624

Text #3

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4702. Min Act: -0.1700

Data Index: 1363110 (Python Code)

Max Activating Token Index: 333

Text #4

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4075. Min Act: -0.1700

Data Index: 480058 (C4 (Web Text))

Max Activating Token Index: 938

Text #5

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4075. Min Act: -0.1700

Data Index: 1381194 (Python Code)

Max Activating Token Index: 1004

Text #6

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4232. Min Act: -0.1700

Data Index: 642041 (C4 (Web Text))

Max Activating Token Index: 929

Text #7

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4232. Min Act: -0.1700

Data Index: 1182456 (C4 (Web Text))

Max Activating Token Index: 284

Text #8

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3918. Min Act: -0.1700

Data Index: 1103074 (C4 (Web Text))

Max Activating Token Index: 241

Text #9

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.4075. Min Act: -0.1700

Data Index: 67705 (C4 (Web Text))

Max Activating Token Index: 46

Text #10

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3761. Min Act: -0.1700

Data Index: 1471817 (Python Code)

Max Activating Token Index: 360

Text #11

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3918. Min Act: -0.1700

Data Index: 1522394 (Python Code)

Max Activating Token Index: 145

Text #12

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3918. Min Act: -0.1700

Data Index: 1227504 (C4 (Web Text))

Max Activating Token Index: 714

Text #13

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3761. Min Act: -0.1700

Data Index: 1449499 (Python Code)

Max Activating Token Index: 942

Text #14

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3291. Min Act: -0.1700

Data Index: 1414041 (Python Code)

Max Activating Token Index: 492

Text #15

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3761. Min Act: -0.1700

Data Index: 760071 (C4 (Web Text))

Max Activating Token Index: 500

Text #16

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3761. Min Act: -0.1700

Data Index: 801515 (C4 (Web Text))

Max Activating Token Index: 967

Text #17

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3918. Min Act: -0.1700

Data Index: 810255 (C4 (Web Text))

Max Activating Token Index: 463

Text #18

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3448. Min Act: -0.1700

Data Index: 195701 (C4 (Web Text))

Max Activating Token Index: 113

Text #19

Max Range: 3.4859. Min Range: -3.4859

Max Act: 3.3605. Min Act: -0.1700

Data Index: 1523913 (Python Code)

Max Activating Token Index: 926