Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 4594 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2728. Min Act: -0.0000

Data Index: 1606954 (Python Code)

Max Activating Token Index: 76

Text #1

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2407. Min Act: -0.0000

Data Index: 1124403 (C4 (Web Text))

Max Activating Token Index: 561

Text #2

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2474. Min Act: -0.0000

Data Index: 1403914 (Python Code)

Max Activating Token Index: 308

Text #3

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2006. Min Act: -0.0000

Data Index: 1620482 (Python Code)

Max Activating Token Index: 293

Text #4

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2115. Min Act: -0.0000

Data Index: 874382 (C4 (Web Text))

Max Activating Token Index: 247

Text #5

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2098. Min Act: -0.0000

Data Index: 1269750 (C4 (Web Text))

Max Activating Token Index: 484

Text #6

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2265. Min Act: -0.0000

Data Index: 1381347 (Python Code)

Max Activating Token Index: 818

Text #7

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2072. Min Act: -0.0000

Data Index: 1306512 (C4 (Web Text))

Max Activating Token Index: 274

Text #8

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2144. Min Act: -0.0000

Data Index: 1291107 (C4 (Web Text))

Max Activating Token Index: 659

Text #9

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2110. Min Act: -0.0000

Data Index: 1612295 (Python Code)

Max Activating Token Index: 401

Text #10

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.1967. Min Act: -0.0000

Data Index: 1611545 (Python Code)

Max Activating Token Index: 374

Text #11

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2187. Min Act: -0.0000

Data Index: 1382350 (Python Code)

Max Activating Token Index: 430

Text #12

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2041. Min Act: -0.0000

Data Index: 83923 (C4 (Web Text))

Max Activating Token Index: 872

Text #13

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2011. Min Act: -0.0000

Data Index: 165173 (C4 (Web Text))

Max Activating Token Index: 601

Text #14

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.1942. Min Act: -0.0000

Data Index: 1002200 (C4 (Web Text))

Max Activating Token Index: 137

Text #15

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.1904. Min Act: -0.0000

Data Index: 1444088 (Python Code)

Max Activating Token Index: 327

Text #16

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2038. Min Act: -0.0000

Data Index: 1199394 (C4 (Web Text))

Max Activating Token Index: 105

Text #17

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.1904. Min Act: -0.0000

Data Index: 1474226 (Python Code)

Max Activating Token Index: 596

Text #18

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.1953. Min Act: -0.0000

Data Index: 1391390 (Python Code)

Max Activating Token Index: 570

Text #19

Max Range: 0.2728. Min Range: -0.2728

Max Act: 0.2093. Min Act: -0.0000

Data Index: 1475117 (Python Code)

Max Activating Token Index: 606