Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 26 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 2.0847. Min Range: -2.0847

Max Act: 2.0257. Min Act: -0.0000

Data Index: 1612643 (Python Code)

Max Activating Token Index: 762

Text #1

Max Range: 2.0847. Min Range: -2.0847

Max Act: 2.0847. Min Act: -0.0000

Data Index: 1486983 (Python Code)

Max Activating Token Index: 337

Text #2

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.8925. Min Act: -0.0000

Data Index: 39337 (C4 (Web Text))

Max Activating Token Index: 498

Text #3

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.8924. Min Act: -0.0000

Data Index: 1553276 (Python Code)

Max Activating Token Index: 739

Text #4

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.4939. Min Act: -0.0000

Data Index: 1243984 (C4 (Web Text))

Max Activating Token Index: 331

Text #5

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.7569. Min Act: -0.0000

Data Index: 1168922 (C4 (Web Text))

Max Activating Token Index: 1007

Text #6

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.7374. Min Act: -0.0000

Data Index: 794285 (C4 (Web Text))

Max Activating Token Index: 966

Text #7

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.7068. Min Act: -0.0000

Data Index: 42755 (C4 (Web Text))

Max Activating Token Index: 489

Text #8

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6048. Min Act: -0.0000

Data Index: 1358666 (C4 (Web Text))

Max Activating Token Index: 964

Text #9

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.7071. Min Act: -0.0000

Data Index: 542155 (C4 (Web Text))

Max Activating Token Index: 378

Text #10

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6893. Min Act: -0.0000

Data Index: 339505 (C4 (Web Text))

Max Activating Token Index: 445

Text #11

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6502. Min Act: -0.0000

Data Index: 1492432 (Python Code)

Max Activating Token Index: 982

Text #12

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.4944. Min Act: -0.0000

Data Index: 1381244 (Python Code)

Max Activating Token Index: 89

Text #13

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6611. Min Act: -0.0000

Data Index: 570510 (C4 (Web Text))

Max Activating Token Index: 907

Text #14

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6646. Min Act: -0.0000

Data Index: 1480932 (Python Code)

Max Activating Token Index: 13

Text #15

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6312. Min Act: -0.0000

Data Index: 1134150 (C4 (Web Text))

Max Activating Token Index: 450

Text #16

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.7050. Min Act: -0.0000

Data Index: 1488644 (Python Code)

Max Activating Token Index: 733

Text #17

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6013. Min Act: -0.0000

Data Index: 1648985 (Python Code)

Max Activating Token Index: 584

Text #18

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.5103. Min Act: -0.0000

Data Index: 934860 (C4 (Web Text))

Max Activating Token Index: 755

Text #19

Max Range: 2.0847. Min Range: -2.0847

Max Act: 1.6465. Min Act: -0.0000

Data Index: 1326561 (C4 (Web Text))

Max Activating Token Index: 522