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

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

Neuron 3230 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0664. Min Act: -0.0001

Data Index: 855003 (C4 (Web Text))

Max Activating Token Index: 469

Text #1

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0663. Min Act: -0.0001

Data Index: 744508 (C4 (Web Text))

Max Activating Token Index: 370

Text #2

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0687. Min Act: -0.0000

Data Index: 582028 (C4 (Web Text))

Max Activating Token Index: 415

Text #3

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0645. Min Act: -0.0000

Data Index: 106561 (C4 (Web Text))

Max Activating Token Index: 588

Text #4

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0659. Min Act: -0.0001

Data Index: 1337609 (C4 (Web Text))

Max Activating Token Index: 262

Text #5

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0593. Min Act: -0.0001

Data Index: 341220 (C4 (Web Text))

Max Activating Token Index: 415

Text #6

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0545. Min Act: -0.0001

Data Index: 230078 (C4 (Web Text))

Max Activating Token Index: 355

Text #7

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0612. Min Act: -0.0000

Data Index: 445290 (C4 (Web Text))

Max Activating Token Index: 233

Text #8

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0516. Min Act: -0.0001

Data Index: 1238508 (C4 (Web Text))

Max Activating Token Index: 203

Text #9

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0497. Min Act: -0.0000

Data Index: 591126 (C4 (Web Text))

Max Activating Token Index: 433

Text #10

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0480. Min Act: -0.0000

Data Index: 382802 (C4 (Web Text))

Max Activating Token Index: 983

Text #11

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0451. Min Act: -0.0001

Data Index: 269584 (C4 (Web Text))

Max Activating Token Index: 412

Text #12

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0415. Min Act: -0.0000

Data Index: 1566077 (Python Code)

Max Activating Token Index: 565

Text #13

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0437. Min Act: -0.0001

Data Index: 1307542 (C4 (Web Text))

Max Activating Token Index: 741

Text #14

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0454. Min Act: -0.0000

Data Index: 498959 (C4 (Web Text))

Max Activating Token Index: 596

Text #15

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0429. Min Act: -0.0000

Data Index: 943120 (C4 (Web Text))

Max Activating Token Index: 333

Text #16

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0430. Min Act: -0.0000

Data Index: 694611 (C4 (Web Text))

Max Activating Token Index: 81

Text #17

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0364. Min Act: -0.0001

Data Index: 303318 (C4 (Web Text))

Max Activating Token Index: 145

Text #18

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0346. Min Act: -0.0001

Data Index: 1265541 (C4 (Web Text))

Max Activating Token Index: 590

Text #19

Max Range: 0.0687. Min Range: -0.0687

Max Act: 0.0387. Min Act: -0.0001

Data Index: 980034 (C4 (Web Text))

Max Activating Token Index: 744