Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 4091 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0835. Min Act: 0.0000

Data Index: 676560 (C4 (Web Text))

Max Activating Token Index: 793

Text #1

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0667. Min Act: -0.0000

Data Index: 1512367 (Python Code)

Max Activating Token Index: 430

Text #2

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0690. Min Act: -0.0001

Data Index: 1045080 (C4 (Web Text))

Max Activating Token Index: 831

Text #3

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0571. Min Act: -0.0001

Data Index: 1402348 (Python Code)

Max Activating Token Index: 715

Text #4

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0592. Min Act: -0.0000

Data Index: 727777 (C4 (Web Text))

Max Activating Token Index: 800

Text #5

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0569. Min Act: -0.0000

Data Index: 187413 (C4 (Web Text))

Max Activating Token Index: 287

Text #6

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0586. Min Act: -0.0001

Data Index: 916588 (C4 (Web Text))

Max Activating Token Index: 1022

Text #7

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0571. Min Act: 0.0001

Data Index: 672982 (C4 (Web Text))

Max Activating Token Index: 142

Text #8

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0549. Min Act: -0.0001

Data Index: 1240614 (C4 (Web Text))

Max Activating Token Index: 794

Text #9

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0588. Min Act: -0.0000

Data Index: 1594785 (Python Code)

Max Activating Token Index: 606

Text #10

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0544. Min Act: -0.0000

Data Index: 193124 (C4 (Web Text))

Max Activating Token Index: 538

Text #11

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0523. Min Act: -0.0001

Data Index: 1515830 (Python Code)

Max Activating Token Index: 813

Text #12

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0541. Min Act: -0.0000

Data Index: 820786 (C4 (Web Text))

Max Activating Token Index: 689

Text #13

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0543. Min Act: -0.0000

Data Index: 320010 (C4 (Web Text))

Max Activating Token Index: 105

Text #14

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0521. Min Act: -0.0000

Data Index: 369388 (C4 (Web Text))

Max Activating Token Index: 662

Text #15

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0530. Min Act: -0.0000

Data Index: 104805 (C4 (Web Text))

Max Activating Token Index: 160

Text #16

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0490. Min Act: -0.0001

Data Index: 1533744 (Python Code)

Max Activating Token Index: 617

Text #17

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0522. Min Act: -0.0000

Data Index: 556469 (C4 (Web Text))

Max Activating Token Index: 659

Text #18

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0501. Min Act: -0.0000

Data Index: 399147 (C4 (Web Text))

Max Activating Token Index: 685

Text #19

Max Range: 0.0835. Min Range: -0.0835

Max Act: 0.0520. Min Act: -0.0001

Data Index: 657150 (C4 (Web Text))

Max Activating Token Index: 526