Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 1876 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.5180. Min Act: -0.0003

Data Index: 1249244 (C4 (Web Text))

Max Activating Token Index: 52

Text #1

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4942. Min Act: -0.0003

Data Index: 1436233 (Python Code)

Max Activating Token Index: 866

Text #2

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.5140. Min Act: -0.0003

Data Index: 1260719 (C4 (Web Text))

Max Activating Token Index: 914

Text #3

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.5086. Min Act: -0.0003

Data Index: 66633 (C4 (Web Text))

Max Activating Token Index: 906

Text #4

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4879. Min Act: -0.0003

Data Index: 1335320 (C4 (Web Text))

Max Activating Token Index: 483

Text #5

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4824. Min Act: -0.0003

Data Index: 1375250 (Python Code)

Max Activating Token Index: 428

Text #6

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4290. Min Act: -0.0003

Data Index: 941406 (C4 (Web Text))

Max Activating Token Index: 199

Text #7

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4591. Min Act: -0.0003

Data Index: 953806 (C4 (Web Text))

Max Activating Token Index: 986

Text #8

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4523. Min Act: -0.0003

Data Index: 1031389 (C4 (Web Text))

Max Activating Token Index: 104

Text #9

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4405. Min Act: -0.0003

Data Index: 82598 (C4 (Web Text))

Max Activating Token Index: 497

Text #10

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4009. Min Act: -0.0003

Data Index: 931232 (C4 (Web Text))

Max Activating Token Index: 530

Text #11

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4092. Min Act: -0.0003

Data Index: 1393336 (Python Code)

Max Activating Token Index: 109

Text #12

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4446. Min Act: -0.0003

Data Index: 401858 (C4 (Web Text))

Max Activating Token Index: 688

Text #13

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4162. Min Act: -0.0003

Data Index: 603689 (C4 (Web Text))

Max Activating Token Index: 850

Text #14

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.3988. Min Act: -0.0003

Data Index: 747330 (C4 (Web Text))

Max Activating Token Index: 86

Text #15

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4097. Min Act: -0.0003

Data Index: 15597 (C4 (Web Text))

Max Activating Token Index: 690

Text #16

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4091. Min Act: -0.0003

Data Index: 1232346 (C4 (Web Text))

Max Activating Token Index: 540

Text #17

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.3802. Min Act: -0.0003

Data Index: 779021 (C4 (Web Text))

Max Activating Token Index: 613

Text #18

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.4073. Min Act: -0.0003

Data Index: 11409 (C4 (Web Text))

Max Activating Token Index: 686

Text #19

Max Range: 0.5180. Min Range: -0.5180

Max Act: 0.3914. Min Act: -0.0003

Data Index: 1121030 (C4 (Web Text))

Max Activating Token Index: 16