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

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

Neuron 2021 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1836. Min Act: -0.0001

Data Index: 816351 (C4 (Web Text))

Max Activating Token Index: 693

Text #1

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1975. Min Act: -0.0001

Data Index: 568821 (C4 (Web Text))

Max Activating Token Index: 714

Text #2

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1682. Min Act: -0.0001

Data Index: 879351 (C4 (Web Text))

Max Activating Token Index: 305

Text #3

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1767. Min Act: -0.0001

Data Index: 1088034 (C4 (Web Text))

Max Activating Token Index: 221

Text #4

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1697. Min Act: -0.0001

Data Index: 683316 (C4 (Web Text))

Max Activating Token Index: 490

Text #5

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1765. Min Act: -0.0001

Data Index: 352789 (C4 (Web Text))

Max Activating Token Index: 627

Text #6

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1672. Min Act: -0.0001

Data Index: 692591 (C4 (Web Text))

Max Activating Token Index: 584

Text #7

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1726. Min Act: -0.0001

Data Index: 172573 (C4 (Web Text))

Max Activating Token Index: 21

Text #8

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1563. Min Act: -0.0001

Data Index: 124397 (C4 (Web Text))

Max Activating Token Index: 443

Text #9

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1621. Min Act: -0.0001

Data Index: 763698 (C4 (Web Text))

Max Activating Token Index: 375

Text #10

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1262. Min Act: -0.0001

Data Index: 1201032 (C4 (Web Text))

Max Activating Token Index: 280

Text #11

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1602. Min Act: -0.0001

Data Index: 882369 (C4 (Web Text))

Max Activating Token Index: 631

Text #12

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1433. Min Act: -0.0001

Data Index: 378164 (C4 (Web Text))

Max Activating Token Index: 244

Text #13

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1572. Min Act: -0.0001

Data Index: 1191525 (C4 (Web Text))

Max Activating Token Index: 374

Text #14

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1525. Min Act: -0.0001

Data Index: 1329364 (C4 (Web Text))

Max Activating Token Index: 979

Text #15

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1430. Min Act: -0.0001

Data Index: 1526072 (Python Code)

Max Activating Token Index: 381

Text #16

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1414. Min Act: -0.0001

Data Index: 520400 (C4 (Web Text))

Max Activating Token Index: 194

Text #17

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1370. Min Act: -0.0001

Data Index: 920700 (C4 (Web Text))

Max Activating Token Index: 1013

Text #18

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1489. Min Act: -0.0001

Data Index: 1160663 (C4 (Web Text))

Max Activating Token Index: 608

Text #19

Max Range: 0.1975. Min Range: -0.1975

Max Act: 0.1373. Min Act: -0.0001

Data Index: 687920 (C4 (Web Text))

Max Activating Token Index: 269