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

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

Neuron 1272 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: 5.7390. Min Range: -5.7390

Max Act: 5.7390. Min Act: -0.1700

Data Index: 73750 (C4 (Web Text))

Max Activating Token Index: 797

Text #1

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.6140. Min Act: -0.1700

Data Index: 1118351 (C4 (Web Text))

Max Activating Token Index: 501

Text #2

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.5828. Min Act: -0.1700

Data Index: 924331 (C4 (Web Text))

Max Activating Token Index: 982

Text #3

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.5203. Min Act: -0.1700

Data Index: 1649690 (Python Code)

Max Activating Token Index: 597

Text #4

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.4578. Min Act: -0.1700

Data Index: 1100027 (C4 (Web Text))

Max Activating Token Index: 370

Text #5

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.5515. Min Act: -0.1700

Data Index: 31685 (C4 (Web Text))

Max Activating Token Index: 595

Text #6

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.4578. Min Act: -0.1700

Data Index: 634102 (C4 (Web Text))

Max Activating Token Index: 639

Text #7

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3640. Min Act: -0.1700

Data Index: 637922 (C4 (Web Text))

Max Activating Token Index: 818

Text #8

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3953. Min Act: -0.1700

Data Index: 336567 (C4 (Web Text))

Max Activating Token Index: 573

Text #9

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3015. Min Act: -0.1700

Data Index: 1438656 (Python Code)

Max Activating Token Index: 353

Text #10

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3953. Min Act: -0.1700

Data Index: 1552556 (Python Code)

Max Activating Token Index: 15

Text #11

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3953. Min Act: -0.1700

Data Index: 1377206 (Python Code)

Max Activating Token Index: 154

Text #12

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3328. Min Act: -0.1700

Data Index: 1254305 (C4 (Web Text))

Max Activating Token Index: 432

Text #13

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3015. Min Act: -0.1700

Data Index: 1348092 (C4 (Web Text))

Max Activating Token Index: 546

Text #14

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3015. Min Act: -0.1700

Data Index: 921478 (C4 (Web Text))

Max Activating Token Index: 630

Text #15

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3015. Min Act: -0.1700

Data Index: 1079250 (C4 (Web Text))

Max Activating Token Index: 488

Text #16

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3328. Min Act: -0.1700

Data Index: 1504265 (Python Code)

Max Activating Token Index: 44

Text #17

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3015. Min Act: -0.1700

Data Index: 1025677 (C4 (Web Text))

Max Activating Token Index: 57

Text #18

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3328. Min Act: -0.1700

Data Index: 355194 (C4 (Web Text))

Max Activating Token Index: 454

Text #19

Max Range: 5.7390. Min Range: -5.7390

Max Act: 5.3328. Min Act: -0.1700

Data Index: 1001337 (C4 (Web Text))

Max Activating Token Index: 729