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

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

Neuron 3275 in Layer 10

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 2.1993. Min Range: -2.1993

Max Act: 2.1993. Min Act: -0.0000

Data Index: 347075 (C4 (Web Text))

Max Activating Token Index: 464

Text #1

Max Range: 2.1993. Min Range: -2.1993

Max Act: 2.0636. Min Act: 0.0000

Data Index: 1263407 (C4 (Web Text))

Max Activating Token Index: 174

Text #2

Max Range: 2.1993. Min Range: -2.1993

Max Act: 2.0677. Min Act: -0.0000

Data Index: 1362721 (Python Code)

Max Activating Token Index: 360

Text #3

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.9139. Min Act: -0.0000

Data Index: 1229504 (C4 (Web Text))

Max Activating Token Index: 269

Text #4

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.8096. Min Act: -0.0000

Data Index: 1092957 (C4 (Web Text))

Max Activating Token Index: 49

Text #5

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.7912. Min Act: -0.0000

Data Index: 1467821 (Python Code)

Max Activating Token Index: 497

Text #6

Max Range: 2.1993. Min Range: -2.1993

Max Act: 2.0072. Min Act: -0.0000

Data Index: 870117 (C4 (Web Text))

Max Activating Token Index: 716

Text #7

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.7366. Min Act: -0.0000

Data Index: 1639221 (Python Code)

Max Activating Token Index: 209

Text #8

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.6772. Min Act: -0.0000

Data Index: 775389 (C4 (Web Text))

Max Activating Token Index: 241

Text #9

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.7662. Min Act: -0.0000

Data Index: 1033998 (C4 (Web Text))

Max Activating Token Index: 541

Text #10

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.7648. Min Act: -0.0000

Data Index: 341455 (C4 (Web Text))

Max Activating Token Index: 473

Text #11

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.7726. Min Act: -0.0000

Data Index: 68008 (C4 (Web Text))

Max Activating Token Index: 521

Text #12

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.5674. Min Act: -0.0000

Data Index: 303447 (C4 (Web Text))

Max Activating Token Index: 130

Text #13

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.7279. Min Act: -0.0000

Data Index: 1314471 (C4 (Web Text))

Max Activating Token Index: 658

Text #14

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.7320. Min Act: -0.0000

Data Index: 477060 (C4 (Web Text))

Max Activating Token Index: 797

Text #15

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.6925. Min Act: -0.0000

Data Index: 1640115 (Python Code)

Max Activating Token Index: 126

Text #16

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.5728. Min Act: -0.0000

Data Index: 226594 (C4 (Web Text))

Max Activating Token Index: 1000

Text #17

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.5997. Min Act: -0.0000

Data Index: 767953 (C4 (Web Text))

Max Activating Token Index: 820

Text #18

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.6358. Min Act: -0.0000

Data Index: 755221 (C4 (Web Text))

Max Activating Token Index: 467

Text #19

Max Range: 2.1993. Min Range: -2.1993

Max Act: 1.8099. Min Act: -0.0000

Data Index: 1010085 (C4 (Web Text))

Max Activating Token Index: 251