Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3518 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0785. Min Act: -0.0003

Data Index: 630923 (C4 (Web Text))

Max Activating Token Index: 31

Text #1

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0771. Min Act: -0.0003

Data Index: 1612399 (Python Code)

Max Activating Token Index: 259

Text #2

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0764. Min Act: -0.0003

Data Index: 42211 (C4 (Web Text))

Max Activating Token Index: 684

Text #3

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0759. Min Act: -0.0003

Data Index: 906822 (C4 (Web Text))

Max Activating Token Index: 678

Text #4

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0732. Min Act: -0.0003

Data Index: 1576922 (Python Code)

Max Activating Token Index: 74

Text #5

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0754. Min Act: -0.0003

Data Index: 1350585 (C4 (Web Text))

Max Activating Token Index: 211

Text #6

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0780. Min Act: -0.0003

Data Index: 1168279 (C4 (Web Text))

Max Activating Token Index: 572

Text #7

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0726. Min Act: -0.0003

Data Index: 1493644 (Python Code)

Max Activating Token Index: 820

Text #8

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0748. Min Act: -0.0003

Data Index: 1146864 (C4 (Web Text))

Max Activating Token Index: 417

Text #9

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0772. Min Act: -0.0003

Data Index: 1240627 (C4 (Web Text))

Max Activating Token Index: 294

Text #10

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0741. Min Act: -0.0003

Data Index: 1209125 (C4 (Web Text))

Max Activating Token Index: 328

Text #11

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0737. Min Act: -0.0003

Data Index: 1558091 (Python Code)

Max Activating Token Index: 135

Text #12

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0736. Min Act: -0.0004

Data Index: 1545620 (Python Code)

Max Activating Token Index: 410

Text #13

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0722. Min Act: -0.0003

Data Index: 760551 (C4 (Web Text))

Max Activating Token Index: 949

Text #14

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0716. Min Act: -0.0003

Data Index: 1162336 (C4 (Web Text))

Max Activating Token Index: 527

Text #15

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0700. Min Act: -0.0003

Data Index: 1159757 (C4 (Web Text))

Max Activating Token Index: 744

Text #16

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0667. Min Act: -0.0003

Data Index: 938618 (C4 (Web Text))

Max Activating Token Index: 334

Text #17

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0705. Min Act: -0.0003

Data Index: 1136190 (C4 (Web Text))

Max Activating Token Index: 580

Text #18

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0692. Min Act: -0.0004

Data Index: 1434806 (Python Code)

Max Activating Token Index: 227

Text #19

Max Range: 0.0785. Min Range: -0.0785

Max Act: 0.0704. Min Act: -0.0003

Data Index: 658664 (C4 (Web Text))

Max Activating Token Index: 448