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

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

Neuron 2560 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: 0.4001. Min Range: -0.4001

Max Act: 0.4001. Min Act: -0.0001

Data Index: 1570281 (Python Code)

Max Activating Token Index: 276

Text #1

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2962. Min Act: -0.0001

Data Index: 534382 (C4 (Web Text))

Max Activating Token Index: 311

Text #2

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2764. Min Act: -0.0001

Data Index: 1126355 (C4 (Web Text))

Max Activating Token Index: 362

Text #3

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2654. Min Act: -0.0001

Data Index: 44986 (C4 (Web Text))

Max Activating Token Index: 903

Text #4

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2665. Min Act: -0.0001

Data Index: 641046 (C4 (Web Text))

Max Activating Token Index: 390

Text #5

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2661. Min Act: -0.0001

Data Index: 558267 (C4 (Web Text))

Max Activating Token Index: 558

Text #6

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2655. Min Act: -0.0001

Data Index: 1483557 (Python Code)

Max Activating Token Index: 641

Text #7

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2631. Min Act: -0.0001

Data Index: 901510 (C4 (Web Text))

Max Activating Token Index: 196

Text #8

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2626. Min Act: -0.0001

Data Index: 1271204 (C4 (Web Text))

Max Activating Token Index: 190

Text #9

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2549. Min Act: -0.0001

Data Index: 1835 (C4 (Web Text))

Max Activating Token Index: 968

Text #10

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2560. Min Act: -0.0001

Data Index: 701527 (C4 (Web Text))

Max Activating Token Index: 979

Text #11

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2648. Min Act: -0.0001

Data Index: 1384775 (Python Code)

Max Activating Token Index: 562

Text #12

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2484. Min Act: -0.0001

Data Index: 534786 (C4 (Web Text))

Max Activating Token Index: 418

Text #13

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2319. Min Act: -0.0001

Data Index: 195406 (C4 (Web Text))

Max Activating Token Index: 408

Text #14

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2551. Min Act: -0.0001

Data Index: 219297 (C4 (Web Text))

Max Activating Token Index: 540

Text #15

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2536. Min Act: -0.0001

Data Index: 749058 (C4 (Web Text))

Max Activating Token Index: 96

Text #16

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2517. Min Act: -0.0001

Data Index: 976947 (C4 (Web Text))

Max Activating Token Index: 164

Text #17

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2568. Min Act: -0.0001

Data Index: 252425 (C4 (Web Text))

Max Activating Token Index: 997

Text #18

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2462. Min Act: -0.0001

Data Index: 614113 (C4 (Web Text))

Max Activating Token Index: 950

Text #19

Max Range: 0.4001. Min Range: -0.4001

Max Act: 0.2531. Min Act: -0.0001

Data Index: 1402159 (Python Code)

Max Activating Token Index: 919