Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 5033 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 2.3288. Min Range: -2.3288

Max Act: 2.3288. Min Act: -0.0000

Data Index: 1404737 (Python Code)

Max Activating Token Index: 754

Text #1

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.8876. Min Act: -0.0000

Data Index: 555682 (C4 (Web Text))

Max Activating Token Index: 259

Text #2

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.9538. Min Act: -0.0000

Data Index: 805907 (C4 (Web Text))

Max Activating Token Index: 49

Text #3

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.9443. Min Act: -0.0000

Data Index: 189118 (C4 (Web Text))

Max Activating Token Index: 584

Text #4

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.7369. Min Act: -0.0000

Data Index: 624329 (C4 (Web Text))

Max Activating Token Index: 840

Text #5

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.7197. Min Act: -0.0000

Data Index: 310481 (C4 (Web Text))

Max Activating Token Index: 973

Text #6

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.5969. Min Act: -0.0000

Data Index: 1645873 (Python Code)

Max Activating Token Index: 71

Text #7

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.6738. Min Act: -0.0000

Data Index: 1276417 (C4 (Web Text))

Max Activating Token Index: 759

Text #8

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.5122. Min Act: -0.0000

Data Index: 259598 (C4 (Web Text))

Max Activating Token Index: 95

Text #9

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.5559. Min Act: -0.0000

Data Index: 825250 (C4 (Web Text))

Max Activating Token Index: 34

Text #10

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4847. Min Act: -0.0000

Data Index: 668194 (C4 (Web Text))

Max Activating Token Index: 721

Text #11

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4872. Min Act: -0.0000

Data Index: 780933 (C4 (Web Text))

Max Activating Token Index: 596

Text #12

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4778. Min Act: -0.0000

Data Index: 1251289 (C4 (Web Text))

Max Activating Token Index: 558

Text #13

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4699. Min Act: -0.0000

Data Index: 602445 (C4 (Web Text))

Max Activating Token Index: 196

Text #14

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4742. Min Act: -0.0000

Data Index: 773457 (C4 (Web Text))

Max Activating Token Index: 478

Text #15

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.3829. Min Act: -0.0000

Data Index: 895981 (C4 (Web Text))

Max Activating Token Index: 857

Text #16

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4689. Min Act: -0.0000

Data Index: 282223 (C4 (Web Text))

Max Activating Token Index: 490

Text #17

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4672. Min Act: -0.0000

Data Index: 256651 (C4 (Web Text))

Max Activating Token Index: 784

Text #18

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.3728. Min Act: -0.0000

Data Index: 869222 (C4 (Web Text))

Max Activating Token Index: 993

Text #19

Max Range: 2.3288. Min Range: -2.3288

Max Act: 1.4656. Min Act: -0.0000

Data Index: 1556679 (Python Code)

Max Activating Token Index: 1023