Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 970 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0922. Min Act: -0.0002

Data Index: 1557028 (Python Code)

Max Activating Token Index: 519

Text #1

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0926. Min Act: -0.0001

Data Index: 1632515 (Python Code)

Max Activating Token Index: 335

Text #2

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0897. Min Act: -0.0002

Data Index: 1504976 (Python Code)

Max Activating Token Index: 374

Text #3

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0846. Min Act: -0.0002

Data Index: 500176 (C4 (Web Text))

Max Activating Token Index: 409

Text #4

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0826. Min Act: -0.0002

Data Index: 545303 (C4 (Web Text))

Max Activating Token Index: 467

Text #5

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0768. Min Act: -0.0002

Data Index: 1342712 (C4 (Web Text))

Max Activating Token Index: 1019

Text #6

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0809. Min Act: -0.0002

Data Index: 58034 (C4 (Web Text))

Max Activating Token Index: 353

Text #7

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0800. Min Act: -0.0002

Data Index: 1293203 (C4 (Web Text))

Max Activating Token Index: 18

Text #8

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0781. Min Act: -0.0002

Data Index: 134578 (C4 (Web Text))

Max Activating Token Index: 396

Text #9

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0773. Min Act: -0.0002

Data Index: 1001883 (C4 (Web Text))

Max Activating Token Index: 228

Text #10

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0777. Min Act: -0.0002

Data Index: 1438284 (Python Code)

Max Activating Token Index: 115

Text #11

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0783. Min Act: -0.0002

Data Index: 1034784 (C4 (Web Text))

Max Activating Token Index: 373

Text #12

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0741. Min Act: -0.0002

Data Index: 388536 (C4 (Web Text))

Max Activating Token Index: 391

Text #13

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0731. Min Act: -0.0002

Data Index: 664888 (C4 (Web Text))

Max Activating Token Index: 310

Text #14

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0730. Min Act: -0.0002

Data Index: 698289 (C4 (Web Text))

Max Activating Token Index: 598

Text #15

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0755. Min Act: -0.0002

Data Index: 17086 (C4 (Web Text))

Max Activating Token Index: 96

Text #16

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0753. Min Act: -0.0002

Data Index: 651079 (C4 (Web Text))

Max Activating Token Index: 404

Text #17

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0674. Min Act: -0.0002

Data Index: 1363989 (Python Code)

Max Activating Token Index: 710

Text #18

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0697. Min Act: -0.0001

Data Index: 944195 (C4 (Web Text))

Max Activating Token Index: 815

Text #19

Max Range: 0.0926. Min Range: -0.0926

Max Act: 0.0705. Min Act: -0.0002

Data Index: 1224906 (C4 (Web Text))

Max Activating Token Index: 633