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

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

Neuron 2000 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: 4.5820. Min Range: -4.5820

Max Act: 4.5820. Min Act: -0.1700

Data Index: 310167 (C4 (Web Text))

Max Activating Token Index: 5

Text #1

Max Range: 4.5820. Min Range: -4.5820

Max Act: 4.1131. Min Act: -0.1700

Data Index: 730225 (C4 (Web Text))

Max Activating Token Index: 615

Text #2

Max Range: 4.5820. Min Range: -4.5820

Max Act: 4.0193. Min Act: -0.1700

Data Index: 360576 (C4 (Web Text))

Max Activating Token Index: 35

Text #3

Max Range: 4.5820. Min Range: -4.5820

Max Act: 4.0193. Min Act: -0.1700

Data Index: 1336229 (C4 (Web Text))

Max Activating Token Index: 975

Text #4

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9881. Min Act: -0.1700

Data Index: 980734 (C4 (Web Text))

Max Activating Token Index: 10

Text #5

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9568. Min Act: -0.1700

Data Index: 532052 (C4 (Web Text))

Max Activating Token Index: 25

Text #6

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9568. Min Act: -0.1700

Data Index: 215436 (C4 (Web Text))

Max Activating Token Index: 155

Text #7

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9881. Min Act: -0.1700

Data Index: 13806 (C4 (Web Text))

Max Activating Token Index: 1016

Text #8

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1700

Data Index: 132168 (C4 (Web Text))

Max Activating Token Index: 16

Text #9

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1700

Data Index: 1123726 (C4 (Web Text))

Max Activating Token Index: 7

Text #10

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1700

Data Index: 47365 (C4 (Web Text))

Max Activating Token Index: 10

Text #11

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1699

Data Index: 1023905 (C4 (Web Text))

Max Activating Token Index: 131

Text #12

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1700

Data Index: 993355 (C4 (Web Text))

Max Activating Token Index: 762

Text #13

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1700

Data Index: 982548 (C4 (Web Text))

Max Activating Token Index: 856

Text #14

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.8943. Min Act: -0.1700

Data Index: 1309582 (C4 (Web Text))

Max Activating Token Index: 620

Text #15

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.8943. Min Act: -0.1700

Data Index: 537228 (C4 (Web Text))

Max Activating Token Index: 510

Text #16

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1700

Data Index: 437558 (C4 (Web Text))

Max Activating Token Index: 137

Text #17

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.9255. Min Act: -0.1700

Data Index: 262386 (C4 (Web Text))

Max Activating Token Index: 452

Text #18

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.8630. Min Act: -0.1700

Data Index: 447673 (C4 (Web Text))

Max Activating Token Index: 21

Text #19

Max Range: 4.5820. Min Range: -4.5820

Max Act: 3.8630. Min Act: -0.1700

Data Index: 909494 (C4 (Web Text))

Max Activating Token Index: 825