Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 4505 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4714. Min Act: -0.0001

Data Index: 1364806 (Python Code)

Max Activating Token Index: 754

Text #1

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.3842. Min Act: -0.0001

Data Index: 824344 (C4 (Web Text))

Max Activating Token Index: 842

Text #2

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4020. Min Act: -0.0001

Data Index: 179212 (C4 (Web Text))

Max Activating Token Index: 367

Text #3

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4249. Min Act: -0.0001

Data Index: 283521 (C4 (Web Text))

Max Activating Token Index: 15

Text #4

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4615. Min Act: -0.0001

Data Index: 347565 (C4 (Web Text))

Max Activating Token Index: 22

Text #5

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4733. Min Act: -0.0001

Data Index: 69034 (C4 (Web Text))

Max Activating Token Index: 258

Text #6

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.3894. Min Act: -0.0001

Data Index: 1646543 (Python Code)

Max Activating Token Index: 325

Text #7

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4360. Min Act: -0.0001

Data Index: 966038 (C4 (Web Text))

Max Activating Token Index: 25

Text #8

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4481. Min Act: -0.0001

Data Index: 174629 (C4 (Web Text))

Max Activating Token Index: 957

Text #9

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4221. Min Act: -0.0001

Data Index: 1035334 (C4 (Web Text))

Max Activating Token Index: 745

Text #10

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4143. Min Act: -0.0001

Data Index: 435511 (C4 (Web Text))

Max Activating Token Index: 666

Text #11

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4021. Min Act: -0.0001

Data Index: 434843 (C4 (Web Text))

Max Activating Token Index: 34

Text #12

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.3868. Min Act: -0.0001

Data Index: 1248388 (C4 (Web Text))

Max Activating Token Index: 209

Text #13

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4035. Min Act: -0.0001

Data Index: 628848 (C4 (Web Text))

Max Activating Token Index: 113

Text #14

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.3917. Min Act: -0.0001

Data Index: 212235 (C4 (Web Text))

Max Activating Token Index: 491

Text #15

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4293. Min Act: -0.0001

Data Index: 1284874 (C4 (Web Text))

Max Activating Token Index: 594

Text #16

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.3636. Min Act: -0.0001

Data Index: 1190927 (C4 (Web Text))

Max Activating Token Index: 219

Text #17

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.4110. Min Act: -0.0001

Data Index: 1054745 (C4 (Web Text))

Max Activating Token Index: 87

Text #18

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.3432. Min Act: -0.0001

Data Index: 1038454 (C4 (Web Text))

Max Activating Token Index: 354

Text #19

Max Range: 0.4733. Min Range: -0.4733

Max Act: 0.3820. Min Act: -0.0001

Data Index: 595761 (C4 (Web Text))

Max Activating Token Index: 19