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

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

Neuron 4064 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: 0.1098. Min Range: -0.1098

Max Act: 0.1080. Min Act: -0.0001

Data Index: 1103726 (C4 (Web Text))

Max Activating Token Index: 481

Text #1

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.1093. Min Act: -0.0001

Data Index: 958164 (C4 (Web Text))

Max Activating Token Index: 124

Text #2

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0711. Min Act: -0.0001

Data Index: 102409 (C4 (Web Text))

Max Activating Token Index: 741

Text #3

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0784. Min Act: -0.0001

Data Index: 496340 (C4 (Web Text))

Max Activating Token Index: 774

Text #4

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.1040. Min Act: -0.0001

Data Index: 759171 (C4 (Web Text))

Max Activating Token Index: 669

Text #5

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.1098. Min Act: -0.0001

Data Index: 502276 (C4 (Web Text))

Max Activating Token Index: 750

Text #6

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0854. Min Act: -0.0001

Data Index: 450578 (C4 (Web Text))

Max Activating Token Index: 268

Text #7

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0760. Min Act: -0.0001

Data Index: 6761 (C4 (Web Text))

Max Activating Token Index: 576

Text #8

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0820. Min Act: -0.0001

Data Index: 1399949 (Python Code)

Max Activating Token Index: 347

Text #9

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0949. Min Act: -0.0001

Data Index: 396727 (C4 (Web Text))

Max Activating Token Index: 263

Text #10

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0697. Min Act: -0.0001

Data Index: 1635535 (Python Code)

Max Activating Token Index: 834

Text #11

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0969. Min Act: -0.0001

Data Index: 1477498 (Python Code)

Max Activating Token Index: 243

Text #12

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0762. Min Act: -0.0001

Data Index: 1395654 (Python Code)

Max Activating Token Index: 333

Text #13

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0737. Min Act: -0.0001

Data Index: 282189 (C4 (Web Text))

Max Activating Token Index: 1008

Text #14

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0635. Min Act: -0.0001

Data Index: 1643453 (Python Code)

Max Activating Token Index: 887

Text #15

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0718. Min Act: -0.0001

Data Index: 1508599 (Python Code)

Max Activating Token Index: 492

Text #16

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0671. Min Act: -0.0001

Data Index: 1106459 (C4 (Web Text))

Max Activating Token Index: 488

Text #17

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0791. Min Act: -0.0001

Data Index: 140002 (C4 (Web Text))

Max Activating Token Index: 147

Text #18

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0626. Min Act: -0.0001

Data Index: 365896 (C4 (Web Text))

Max Activating Token Index: 826

Text #19

Max Range: 0.1098. Min Range: -0.1098

Max Act: 0.0695. Min Act: -0.0001

Data Index: 90335 (C4 (Web Text))

Max Activating Token Index: 779