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

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

Neuron 4608 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2690. Min Act: -0.0000

Data Index: 207591 (C4 (Web Text))

Max Activating Token Index: 828

Text #1

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.3044. Min Act: -0.0000

Data Index: 1517480 (Python Code)

Max Activating Token Index: 59

Text #2

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2897. Min Act: -0.0000

Data Index: 1363890 (Python Code)

Max Activating Token Index: 39

Text #3

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.3058. Min Act: -0.0000

Data Index: 541161 (C4 (Web Text))

Max Activating Token Index: 611

Text #4

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2724. Min Act: -0.0000

Data Index: 1636873 (Python Code)

Max Activating Token Index: 265

Text #5

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2923. Min Act: -0.0000

Data Index: 1547924 (Python Code)

Max Activating Token Index: 60

Text #6

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2929. Min Act: -0.0000

Data Index: 944291 (C4 (Web Text))

Max Activating Token Index: 984

Text #7

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.3010. Min Act: -0.0000

Data Index: 236126 (C4 (Web Text))

Max Activating Token Index: 1003

Text #8

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2923. Min Act: -0.0000

Data Index: 264696 (C4 (Web Text))

Max Activating Token Index: 1012

Text #9

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2614. Min Act: -0.0000

Data Index: 105817 (C4 (Web Text))

Max Activating Token Index: 966

Text #10

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2746. Min Act: -0.0000

Data Index: 1654869 (Python Code)

Max Activating Token Index: 372

Text #11

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2836. Min Act: -0.0000

Data Index: 1132487 (C4 (Web Text))

Max Activating Token Index: 993

Text #12

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2753. Min Act: -0.0000

Data Index: 1370974 (Python Code)

Max Activating Token Index: 550

Text #13

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2689. Min Act: -0.0000

Data Index: 1607548 (Python Code)

Max Activating Token Index: 635

Text #14

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2947. Min Act: -0.0000

Data Index: 1647725 (Python Code)

Max Activating Token Index: 74

Text #15

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2844. Min Act: -0.0000

Data Index: 1087243 (C4 (Web Text))

Max Activating Token Index: 938

Text #16

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2845. Min Act: -0.0000

Data Index: 1391489 (Python Code)

Max Activating Token Index: 442

Text #17

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2555. Min Act: -0.0000

Data Index: 1559395 (Python Code)

Max Activating Token Index: 768

Text #18

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2636. Min Act: -0.0000

Data Index: 1562588 (Python Code)

Max Activating Token Index: 433

Text #19

Max Range: 0.3058. Min Range: -0.3058

Max Act: 0.2438. Min Act: -0.0000

Data Index: 1235431 (C4 (Web Text))

Max Activating Token Index: 542