Model: GELU Model: 2 Layers, 2048 Neurons per Layer

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

Neuron 900 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-2l')

Text #0

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.9515. Min Act: -0.1700

Data Index: 1263072 (C4 (Web Text))

Max Activating Token Index: 18

Text #1

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.7953. Min Act: -0.1700

Data Index: 163501 (C4 (Web Text))

Max Activating Token Index: 881

Text #2

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.6078. Min Act: -0.1700

Data Index: 512881 (C4 (Web Text))

Max Activating Token Index: 928

Text #3

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.6234. Min Act: -0.1700

Data Index: 1038594 (C4 (Web Text))

Max Activating Token Index: 726

Text #4

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.3890. Min Act: -0.1700

Data Index: 1127255 (C4 (Web Text))

Max Activating Token Index: 866

Text #5

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.4672. Min Act: -0.1700

Data Index: 628552 (C4 (Web Text))

Max Activating Token Index: 723

Text #6

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.3422. Min Act: -0.1700

Data Index: 1647099 (Python Code)

Max Activating Token Index: 1004

Text #7

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.2797. Min Act: -0.1700

Data Index: 1569981 (Python Code)

Max Activating Token Index: 252

Text #8

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.2484. Min Act: -0.1700

Data Index: 1498862 (Python Code)

Max Activating Token Index: 782

Text #9

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.2797. Min Act: -0.1700

Data Index: 201532 (C4 (Web Text))

Max Activating Token Index: 314

Text #10

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.2484. Min Act: -0.1700

Data Index: 836246 (C4 (Web Text))

Max Activating Token Index: 472

Text #11

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.2328. Min Act: -0.1700

Data Index: 1140060 (C4 (Web Text))

Max Activating Token Index: 287

Text #12

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.2172. Min Act: -0.1700

Data Index: 1104773 (C4 (Web Text))

Max Activating Token Index: 46

Text #13

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.0765. Min Act: -0.1700

Data Index: 64254 (C4 (Web Text))

Max Activating Token Index: 867

Text #14

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.1390. Min Act: -0.1700

Data Index: 325662 (C4 (Web Text))

Max Activating Token Index: 187

Text #15

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.1234. Min Act: -0.1700

Data Index: 353542 (C4 (Web Text))

Max Activating Token Index: 811

Text #16

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.0765. Min Act: -0.1700

Data Index: 1593752 (Python Code)

Max Activating Token Index: 928

Text #17

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.0765. Min Act: -0.1700

Data Index: 1320296 (C4 (Web Text))

Max Activating Token Index: 697

Text #18

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.0140. Min Act: -0.1700

Data Index: 111867 (C4 (Web Text))

Max Activating Token Index: 939

Text #19

Max Range: 5.9515. Min Range: -5.9515

Max Act: 5.0297. Min Act: -0.1700

Data Index: 1411287 (Python Code)

Max Activating Token Index: 156