Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 1838 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0688. Min Act: -0.0001

Data Index: 937656 (C4 (Web Text))

Max Activating Token Index: 336

Text #1

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0651. Min Act: -0.0001

Data Index: 1014189 (C4 (Web Text))

Max Activating Token Index: 50

Text #2

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0624. Min Act: -0.0001

Data Index: 452045 (C4 (Web Text))

Max Activating Token Index: 739

Text #3

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0606. Min Act: -0.0001

Data Index: 494836 (C4 (Web Text))

Max Activating Token Index: 439

Text #4

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0655. Min Act: -0.0001

Data Index: 1523228 (Python Code)

Max Activating Token Index: 966

Text #5

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0598. Min Act: -0.0001

Data Index: 1143256 (C4 (Web Text))

Max Activating Token Index: 64

Text #6

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0595. Min Act: -0.0001

Data Index: 490916 (C4 (Web Text))

Max Activating Token Index: 941

Text #7

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0612. Min Act: -0.0001

Data Index: 1296547 (C4 (Web Text))

Max Activating Token Index: 14

Text #8

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0603. Min Act: -0.0001

Data Index: 272531 (C4 (Web Text))

Max Activating Token Index: 1

Text #9

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0599. Min Act: -0.0001

Data Index: 201240 (C4 (Web Text))

Max Activating Token Index: 980

Text #10

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0601. Min Act: -0.0001

Data Index: 452873 (C4 (Web Text))

Max Activating Token Index: 28

Text #11

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0590. Min Act: -0.0001

Data Index: 1054990 (C4 (Web Text))

Max Activating Token Index: 806

Text #12

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0556. Min Act: -0.0001

Data Index: 621707 (C4 (Web Text))

Max Activating Token Index: 875

Text #13

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0570. Min Act: -0.0001

Data Index: 812630 (C4 (Web Text))

Max Activating Token Index: 177

Text #14

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0593. Min Act: -0.0001

Data Index: 97644 (C4 (Web Text))

Max Activating Token Index: 1023

Text #15

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0585. Min Act: -0.0001

Data Index: 1486500 (Python Code)

Max Activating Token Index: 399

Text #16

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0569. Min Act: -0.0001

Data Index: 1498424 (Python Code)

Max Activating Token Index: 131

Text #17

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0585. Min Act: -0.0001

Data Index: 971324 (C4 (Web Text))

Max Activating Token Index: 1

Text #18

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0585. Min Act: -0.0001

Data Index: 1172315 (C4 (Web Text))

Max Activating Token Index: 1

Text #19

Max Range: 0.0688. Min Range: -0.0688

Max Act: 0.0585. Min Act: -0.0001

Data Index: 309865 (C4 (Web Text))

Max Activating Token Index: 1