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

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

Neuron 1994 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: 5.6958. Min Range: -5.6958

Max Act: 5.6958. Min Act: -0.1700

Data Index: 1136719 (C4 (Web Text))

Max Activating Token Index: 11

Text #1

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.2895. Min Act: -0.1700

Data Index: 588909 (C4 (Web Text))

Max Activating Token Index: 77

Text #2

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.2270. Min Act: -0.1700

Data Index: 228233 (C4 (Web Text))

Max Activating Token Index: 760

Text #3

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.1645. Min Act: -0.1700

Data Index: 582108 (C4 (Web Text))

Max Activating Token Index: 77

Text #4

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.1333. Min Act: -0.1700

Data Index: 9445 (C4 (Web Text))

Max Activating Token Index: 46

Text #5

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.1020. Min Act: -0.1700

Data Index: 1015990 (C4 (Web Text))

Max Activating Token Index: 207

Text #6

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.0083. Min Act: -0.1700

Data Index: 392097 (C4 (Web Text))

Max Activating Token Index: 182

Text #7

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9770. Min Act: -0.1700

Data Index: 318913 (C4 (Web Text))

Max Activating Token Index: 580

Text #8

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.1020. Min Act: -0.1700

Data Index: 543423 (C4 (Web Text))

Max Activating Token Index: 729

Text #9

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.0395. Min Act: -0.1700

Data Index: 935084 (C4 (Web Text))

Max Activating Token Index: 84

Text #10

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.0395. Min Act: -0.1700

Data Index: 691873 (C4 (Web Text))

Max Activating Token Index: 469

Text #11

Max Range: 5.6958. Min Range: -5.6958

Max Act: 5.0083. Min Act: -0.1700

Data Index: 260886 (C4 (Web Text))

Max Activating Token Index: 522

Text #12

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9770. Min Act: -0.1700

Data Index: 1457699 (Python Code)

Max Activating Token Index: 80

Text #13

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9458. Min Act: -0.1700

Data Index: 930682 (C4 (Web Text))

Max Activating Token Index: 69

Text #14

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9145. Min Act: -0.1700

Data Index: 560141 (C4 (Web Text))

Max Activating Token Index: 67

Text #15

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9458. Min Act: -0.1700

Data Index: 6375 (C4 (Web Text))

Max Activating Token Index: 7

Text #16

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9770. Min Act: -0.1700

Data Index: 497718 (C4 (Web Text))

Max Activating Token Index: 435

Text #17

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9770. Min Act: -0.1700

Data Index: 1317187 (C4 (Web Text))

Max Activating Token Index: 533

Text #18

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9458. Min Act: -0.1700

Data Index: 531814 (C4 (Web Text))

Max Activating Token Index: 41

Text #19

Max Range: 5.6958. Min Range: -5.6958

Max Act: 4.9458. Min Act: -0.1700

Data Index: 591819 (C4 (Web Text))

Max Activating Token Index: 468