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

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

Neuron 38 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: 3.6676. Min Range: -3.6676

Max Act: 3.6206. Min Act: -0.1700

Data Index: 469450 (C4 (Web Text))

Max Activating Token Index: 825

Text #1

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.6676. Min Act: -0.1700

Data Index: 157503 (C4 (Web Text))

Max Activating Token Index: 50

Text #2

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.6206. Min Act: -0.1700

Data Index: 432681 (C4 (Web Text))

Max Activating Token Index: 579

Text #3

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.6676. Min Act: -0.1700

Data Index: 591939 (C4 (Web Text))

Max Activating Token Index: 839

Text #4

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5737. Min Act: -0.1700

Data Index: 440453 (C4 (Web Text))

Max Activating Token Index: 657

Text #5

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.6206. Min Act: -0.1700

Data Index: 443552 (C4 (Web Text))

Max Activating Token Index: 742

Text #6

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.6050. Min Act: -0.1700

Data Index: 882509 (C4 (Web Text))

Max Activating Token Index: 986

Text #7

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5580. Min Act: -0.1700

Data Index: 801916 (C4 (Web Text))

Max Activating Token Index: 880

Text #8

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5893. Min Act: -0.1700

Data Index: 1485377 (Python Code)

Max Activating Token Index: 387

Text #9

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5737. Min Act: -0.1700

Data Index: 1375904 (Python Code)

Max Activating Token Index: 739

Text #10

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5893. Min Act: -0.1700

Data Index: 1177619 (C4 (Web Text))

Max Activating Token Index: 152

Text #11

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5580. Min Act: -0.1700

Data Index: 431601 (C4 (Web Text))

Max Activating Token Index: 417

Text #12

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5580. Min Act: -0.1700

Data Index: 1137681 (C4 (Web Text))

Max Activating Token Index: 928

Text #13

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5423. Min Act: -0.1700

Data Index: 1263774 (C4 (Web Text))

Max Activating Token Index: 224

Text #14

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5580. Min Act: -0.1700

Data Index: 816468 (C4 (Web Text))

Max Activating Token Index: 102

Text #15

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5267. Min Act: -0.1700

Data Index: 1283193 (C4 (Web Text))

Max Activating Token Index: 262

Text #16

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.4797. Min Act: -0.1700

Data Index: 995706 (C4 (Web Text))

Max Activating Token Index: 976

Text #17

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5267. Min Act: -0.1700

Data Index: 504984 (C4 (Web Text))

Max Activating Token Index: 859

Text #18

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5110. Min Act: -0.1700

Data Index: 1559798 (Python Code)

Max Activating Token Index: 670

Text #19

Max Range: 3.6676. Min Range: -3.6676

Max Act: 3.5267. Min Act: -0.1700

Data Index: 1206930 (C4 (Web Text))

Max Activating Token Index: 457