Model: GELU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 981 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.6602. Min Act: -0.1700

Data Index: 511427 (C4 (Web Text))

Max Activating Token Index: 18

Text #1

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.6445. Min Act: -0.1700

Data Index: 400177 (C4 (Web Text))

Max Activating Token Index: 715

Text #2

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.6132. Min Act: -0.1700

Data Index: 1200980 (C4 (Web Text))

Max Activating Token Index: 52

Text #3

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5506. Min Act: -0.1700

Data Index: 409157 (C4 (Web Text))

Max Activating Token Index: 580

Text #4

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5506. Min Act: -0.1700

Data Index: 691984 (C4 (Web Text))

Max Activating Token Index: 666

Text #5

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5506. Min Act: -0.1700

Data Index: 1606178 (Python Code)

Max Activating Token Index: 339

Text #6

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5506. Min Act: -0.1700

Data Index: 135068 (C4 (Web Text))

Max Activating Token Index: 55

Text #7

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5349. Min Act: -0.1700

Data Index: 1061448 (C4 (Web Text))

Max Activating Token Index: 17

Text #8

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5349. Min Act: -0.1700

Data Index: 994909 (C4 (Web Text))

Max Activating Token Index: 69

Text #9

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5506. Min Act: -0.1700

Data Index: 492233 (C4 (Web Text))

Max Activating Token Index: 180

Text #10

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5192. Min Act: -0.1700

Data Index: 921712 (C4 (Web Text))

Max Activating Token Index: 892

Text #11

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5192. Min Act: -0.1700

Data Index: 347354 (C4 (Web Text))

Max Activating Token Index: 287

Text #12

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5349. Min Act: -0.1700

Data Index: 1471752 (Python Code)

Max Activating Token Index: 462

Text #13

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5036. Min Act: -0.1700

Data Index: 640166 (C4 (Web Text))

Max Activating Token Index: 41

Text #14

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.5192. Min Act: -0.1700

Data Index: 35914 (C4 (Web Text))

Max Activating Token Index: 266

Text #15

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.4879. Min Act: -0.1700

Data Index: 1415983 (Python Code)

Max Activating Token Index: 491

Text #16

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.4879. Min Act: -0.1700

Data Index: 1563020 (Python Code)

Max Activating Token Index: 606

Text #17

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.4722. Min Act: -0.1700

Data Index: 1370147 (Python Code)

Max Activating Token Index: 508

Text #18

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.4722. Min Act: -0.1700

Data Index: 339183 (C4 (Web Text))

Max Activating Token Index: 727

Text #19

Max Range: 3.6602. Min Range: -3.6602

Max Act: 3.4722. Min Act: -0.1700

Data Index: 1132091 (C4 (Web Text))

Max Activating Token Index: 391