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

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

Neuron 450 in Layer 11

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.5700. Min Act: -0.0000

Data Index: 371128 (C4 (Web Text))

Max Activating Token Index: 155

Text #1

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.5138. Min Act: -0.0000

Data Index: 299631 (C4 (Web Text))

Max Activating Token Index: 128

Text #2

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.5017. Min Act: -0.0000

Data Index: 8937 (C4 (Web Text))

Max Activating Token Index: 331

Text #3

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.4726. Min Act: -0.0000

Data Index: 245504 (C4 (Web Text))

Max Activating Token Index: 982

Text #4

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.4121. Min Act: -0.0000

Data Index: 740900 (C4 (Web Text))

Max Activating Token Index: 854

Text #5

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.3986. Min Act: -0.0000

Data Index: 223920 (C4 (Web Text))

Max Activating Token Index: 777

Text #6

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.3258. Min Act: -0.0000

Data Index: 813145 (C4 (Web Text))

Max Activating Token Index: 684

Text #7

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.3258. Min Act: -0.0000

Data Index: 1317328 (C4 (Web Text))

Max Activating Token Index: 466

Text #8

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.3775. Min Act: -0.0000

Data Index: 77647 (C4 (Web Text))

Max Activating Token Index: 511

Text #9

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.3229. Min Act: -0.0000

Data Index: 516109 (C4 (Web Text))

Max Activating Token Index: 488

Text #10

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.3106. Min Act: -0.0000

Data Index: 1628845 (Python Code)

Max Activating Token Index: 627

Text #11

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2987. Min Act: -0.0000

Data Index: 1452598 (Python Code)

Max Activating Token Index: 857

Text #12

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2941. Min Act: -0.0000

Data Index: 1282987 (C4 (Web Text))

Max Activating Token Index: 647

Text #13

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2918. Min Act: -0.0000

Data Index: 777047 (C4 (Web Text))

Max Activating Token Index: 930

Text #14

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2879. Min Act: -0.0000

Data Index: 1368153 (Python Code)

Max Activating Token Index: 549

Text #15

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2950. Min Act: -0.0000

Data Index: 1494568 (Python Code)

Max Activating Token Index: 562

Text #16

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2902. Min Act: -0.0000

Data Index: 1485473 (Python Code)

Max Activating Token Index: 469

Text #17

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2858. Min Act: -0.0000

Data Index: 1069275 (C4 (Web Text))

Max Activating Token Index: 332

Text #18

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2801. Min Act: -0.0000

Data Index: 443840 (C4 (Web Text))

Max Activating Token Index: 384

Text #19

Max Range: 0.5700. Min Range: -0.5700

Max Act: 0.2529. Min Act: -0.0000

Data Index: 1282053 (C4 (Web Text))

Max Activating Token Index: 683