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

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

Neuron 65 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: 4.0067. Min Range: -4.0067

Max Act: 4.0067. Min Act: -0.1700

Data Index: 1534605 (Python Code)

Max Activating Token Index: 283

Text #1

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.9442. Min Act: -0.1700

Data Index: 1174064 (C4 (Web Text))

Max Activating Token Index: 46

Text #2

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.9129. Min Act: -0.1700

Data Index: 202513 (C4 (Web Text))

Max Activating Token Index: 686

Text #3

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.8191. Min Act: -0.1700

Data Index: 1245651 (C4 (Web Text))

Max Activating Token Index: 860

Text #4

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7878. Min Act: -0.1700

Data Index: 1362637 (Python Code)

Max Activating Token Index: 585

Text #5

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7878. Min Act: -0.1700

Data Index: 614386 (C4 (Web Text))

Max Activating Token Index: 586

Text #6

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.8191. Min Act: -0.1700

Data Index: 1173651 (C4 (Web Text))

Max Activating Token Index: 704

Text #7

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.8504. Min Act: -0.1700

Data Index: 226957 (C4 (Web Text))

Max Activating Token Index: 963

Text #8

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7878. Min Act: -0.1700

Data Index: 1188428 (C4 (Web Text))

Max Activating Token Index: 631

Text #9

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.8191. Min Act: -0.1700

Data Index: 484944 (C4 (Web Text))

Max Activating Token Index: 56

Text #10

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7878. Min Act: -0.1700

Data Index: 12609 (C4 (Web Text))

Max Activating Token Index: 45

Text #11

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7252. Min Act: -0.1700

Data Index: 875617 (C4 (Web Text))

Max Activating Token Index: 470

Text #12

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7565. Min Act: -0.1700

Data Index: 344871 (C4 (Web Text))

Max Activating Token Index: 704

Text #13

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7252. Min Act: -0.1700

Data Index: 909420 (C4 (Web Text))

Max Activating Token Index: 903

Text #14

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7565. Min Act: -0.1700

Data Index: 1629386 (Python Code)

Max Activating Token Index: 879

Text #15

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.6939. Min Act: -0.1700

Data Index: 1526291 (Python Code)

Max Activating Token Index: 1023

Text #16

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7565. Min Act: -0.1700

Data Index: 1420156 (Python Code)

Max Activating Token Index: 37

Text #17

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.6939. Min Act: -0.1700

Data Index: 579092 (C4 (Web Text))

Max Activating Token Index: 836

Text #18

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7252. Min Act: -0.1700

Data Index: 1276996 (C4 (Web Text))

Max Activating Token Index: 581

Text #19

Max Range: 4.0067. Min Range: -4.0067

Max Act: 3.7252. Min Act: -0.1700

Data Index: 177310 (C4 (Web Text))

Max Activating Token Index: 77