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

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

Neuron 1664 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 5.1909. Min Range: -5.1909

Max Act: 5.1909. Min Act: -0.1700

Data Index: 1024464 (C4 (Web Text))

Max Activating Token Index: 32

Text #1

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.9722. Min Act: -0.1700

Data Index: 286134 (C4 (Web Text))

Max Activating Token Index: 47

Text #2

Max Range: 5.1909. Min Range: -5.1909

Max Act: 5.0034. Min Act: -0.1700

Data Index: 813373 (C4 (Web Text))

Max Activating Token Index: 242

Text #3

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.8472. Min Act: -0.1700

Data Index: 555363 (C4 (Web Text))

Max Activating Token Index: 36

Text #4

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7847. Min Act: -0.1700

Data Index: 1632649 (Python Code)

Max Activating Token Index: 674

Text #5

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.8159. Min Act: -0.1700

Data Index: 474380 (C4 (Web Text))

Max Activating Token Index: 543

Text #6

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.8159. Min Act: -0.1700

Data Index: 297043 (C4 (Web Text))

Max Activating Token Index: 671

Text #7

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.8159. Min Act: -0.1700

Data Index: 338860 (C4 (Web Text))

Max Activating Token Index: 536

Text #8

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.8159. Min Act: -0.1700

Data Index: 1608775 (Python Code)

Max Activating Token Index: 370

Text #9

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7534. Min Act: -0.1700

Data Index: 706803 (C4 (Web Text))

Max Activating Token Index: 12

Text #10

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7847. Min Act: -0.1700

Data Index: 450083 (C4 (Web Text))

Max Activating Token Index: 3

Text #11

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7534. Min Act: -0.1700

Data Index: 1260631 (C4 (Web Text))

Max Activating Token Index: 898

Text #12

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7221. Min Act: -0.1700

Data Index: 1554046 (Python Code)

Max Activating Token Index: 247

Text #13

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7534. Min Act: -0.1700

Data Index: 685058 (C4 (Web Text))

Max Activating Token Index: 304

Text #14

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7534. Min Act: -0.1700

Data Index: 1355576 (C4 (Web Text))

Max Activating Token Index: 980

Text #15

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7221. Min Act: -0.1700

Data Index: 1305793 (C4 (Web Text))

Max Activating Token Index: 61

Text #16

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7221. Min Act: -0.1700

Data Index: 374727 (C4 (Web Text))

Max Activating Token Index: 152

Text #17

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7534. Min Act: -0.1700

Data Index: 782848 (C4 (Web Text))

Max Activating Token Index: 194

Text #18

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7534. Min Act: -0.1700

Data Index: 1275589 (C4 (Web Text))

Max Activating Token Index: 306

Text #19

Max Range: 5.1909. Min Range: -5.1909

Max Act: 4.7221. Min Act: -0.1700

Data Index: 865942 (C4 (Web Text))

Max Activating Token Index: 133