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

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

Neuron 5731 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.8794. Min Act: -0.0000

Data Index: 1566743 (Python Code)

Max Activating Token Index: 10

Text #1

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.7569. Min Act: -0.0000

Data Index: 577863 (C4 (Web Text))

Max Activating Token Index: 218

Text #2

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6828. Min Act: -0.0000

Data Index: 1073058 (C4 (Web Text))

Max Activating Token Index: 1001

Text #3

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.7817. Min Act: -0.0000

Data Index: 815929 (C4 (Web Text))

Max Activating Token Index: 210

Text #4

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.7314. Min Act: -0.0000

Data Index: 284767 (C4 (Web Text))

Max Activating Token Index: 58

Text #5

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6778. Min Act: -0.0000

Data Index: 150167 (C4 (Web Text))

Max Activating Token Index: 881

Text #6

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.7410. Min Act: -0.0001

Data Index: 31012 (C4 (Web Text))

Max Activating Token Index: 152

Text #7

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6903. Min Act: -0.0000

Data Index: 1298199 (C4 (Web Text))

Max Activating Token Index: 489

Text #8

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.7312. Min Act: -0.0001

Data Index: 1185399 (C4 (Web Text))

Max Activating Token Index: 375

Text #9

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.7020. Min Act: -0.0000

Data Index: 310548 (C4 (Web Text))

Max Activating Token Index: 582

Text #10

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6942. Min Act: -0.0000

Data Index: 820040 (C4 (Web Text))

Max Activating Token Index: 103

Text #11

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6764. Min Act: -0.0000

Data Index: 1464410 (Python Code)

Max Activating Token Index: 45

Text #12

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6520. Min Act: -0.0000

Data Index: 697582 (C4 (Web Text))

Max Activating Token Index: 942

Text #13

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6484. Min Act: -0.0000

Data Index: 1475468 (Python Code)

Max Activating Token Index: 623

Text #14

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6670. Min Act: -0.0000

Data Index: 974813 (C4 (Web Text))

Max Activating Token Index: 641

Text #15

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6436. Min Act: -0.0001

Data Index: 19737 (C4 (Web Text))

Max Activating Token Index: 561

Text #16

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6418. Min Act: -0.0000

Data Index: 220037 (C4 (Web Text))

Max Activating Token Index: 338

Text #17

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6284. Min Act: -0.0000

Data Index: 1127840 (C4 (Web Text))

Max Activating Token Index: 913

Text #18

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6431. Min Act: -0.0000

Data Index: 1564311 (Python Code)

Max Activating Token Index: 730

Text #19

Max Range: 0.8794. Min Range: -0.8794

Max Act: 0.6770. Min Act: -0.0000

Data Index: 1266543 (C4 (Web Text))

Max Activating Token Index: 901