Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 3940 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 3.0761. Min Range: -3.0761

Max Act: 3.0761. Min Act: -0.0001

Data Index: 131486 (C4 (Web Text))

Max Activating Token Index: 157

Text #1

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.7090. Min Act: -0.0001

Data Index: 274152 (C4 (Web Text))

Max Activating Token Index: 313

Text #2

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.5555. Min Act: -0.0001

Data Index: 294959 (C4 (Web Text))

Max Activating Token Index: 178

Text #3

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.6464. Min Act: -0.0001

Data Index: 120225 (C4 (Web Text))

Max Activating Token Index: 853

Text #4

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.4106. Min Act: -0.0001

Data Index: 1646859 (Python Code)

Max Activating Token Index: 164

Text #5

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.3205. Min Act: -0.0001

Data Index: 691199 (C4 (Web Text))

Max Activating Token Index: 767

Text #6

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.5892. Min Act: -0.0001

Data Index: 1618150 (Python Code)

Max Activating Token Index: 759

Text #7

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.5225. Min Act: -0.0001

Data Index: 1584548 (Python Code)

Max Activating Token Index: 832

Text #8

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.4919. Min Act: -0.0001

Data Index: 663284 (C4 (Web Text))

Max Activating Token Index: 506

Text #9

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.3314. Min Act: -0.0001

Data Index: 792087 (C4 (Web Text))

Max Activating Token Index: 940

Text #10

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.3244. Min Act: -0.0001

Data Index: 306875 (C4 (Web Text))

Max Activating Token Index: 416

Text #11

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.4360. Min Act: -0.0001

Data Index: 67508 (C4 (Web Text))

Max Activating Token Index: 589

Text #12

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.3160. Min Act: -0.0001

Data Index: 988102 (C4 (Web Text))

Max Activating Token Index: 711

Text #13

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.1773. Min Act: -0.0001

Data Index: 1412791 (Python Code)

Max Activating Token Index: 77

Text #14

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.2554. Min Act: -0.0000

Data Index: 300127 (C4 (Web Text))

Max Activating Token Index: 127

Text #15

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.5063. Min Act: -0.0000

Data Index: 291073 (C4 (Web Text))

Max Activating Token Index: 476

Text #16

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.4977. Min Act: -0.0001

Data Index: 437934 (C4 (Web Text))

Max Activating Token Index: 788

Text #17

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.1198. Min Act: -0.0000

Data Index: 1237130 (C4 (Web Text))

Max Activating Token Index: 89

Text #18

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.3616. Min Act: -0.0001

Data Index: 791598 (C4 (Web Text))

Max Activating Token Index: 82

Text #19

Max Range: 3.0761. Min Range: -3.0761

Max Act: 2.2408. Min Act: -0.0000

Data Index: 648749 (C4 (Web Text))

Max Activating Token Index: 897