Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 1984 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.7973. Min Act: -0.0003

Data Index: 105828 (C4 (Web Text))

Max Activating Token Index: 607

Text #1

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.7080. Min Act: -0.0003

Data Index: 437747 (C4 (Web Text))

Max Activating Token Index: 523

Text #2

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.6003. Min Act: -0.0003

Data Index: 85488 (C4 (Web Text))

Max Activating Token Index: 346

Text #3

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.6037. Min Act: -0.0003

Data Index: 1606549 (Python Code)

Max Activating Token Index: 898

Text #4

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5940. Min Act: -0.0003

Data Index: 445899 (C4 (Web Text))

Max Activating Token Index: 640

Text #5

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5904. Min Act: -0.0003

Data Index: 611505 (C4 (Web Text))

Max Activating Token Index: 149

Text #6

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5806. Min Act: -0.0003

Data Index: 982373 (C4 (Web Text))

Max Activating Token Index: 832

Text #7

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5941. Min Act: -0.0003

Data Index: 318202 (C4 (Web Text))

Max Activating Token Index: 953

Text #8

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5818. Min Act: -0.0003

Data Index: 674898 (C4 (Web Text))

Max Activating Token Index: 806

Text #9

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5830. Min Act: -0.0003

Data Index: 503422 (C4 (Web Text))

Max Activating Token Index: 258

Text #10

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5435. Min Act: -0.0003

Data Index: 485736 (C4 (Web Text))

Max Activating Token Index: 410

Text #11

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5571. Min Act: -0.0003

Data Index: 829578 (C4 (Web Text))

Max Activating Token Index: 748

Text #12

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5559. Min Act: -0.0003

Data Index: 925055 (C4 (Web Text))

Max Activating Token Index: 522

Text #13

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5062. Min Act: -0.0003

Data Index: 1239649 (C4 (Web Text))

Max Activating Token Index: 455

Text #14

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5392. Min Act: -0.0002

Data Index: 1651821 (Python Code)

Max Activating Token Index: 684

Text #15

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5356. Min Act: -0.0003

Data Index: 864600 (C4 (Web Text))

Max Activating Token Index: 395

Text #16

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5653. Min Act: -0.0003

Data Index: 771681 (C4 (Web Text))

Max Activating Token Index: 790

Text #17

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5428. Min Act: -0.0003

Data Index: 330203 (C4 (Web Text))

Max Activating Token Index: 1022

Text #18

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5443. Min Act: -0.0003

Data Index: 455079 (C4 (Web Text))

Max Activating Token Index: 395

Text #19

Max Range: 0.7973. Min Range: -0.7973

Max Act: 0.5395. Min Act: -0.0003

Data Index: 646135 (C4 (Web Text))

Max Activating Token Index: 233