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

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

Neuron 3448 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3923. Min Act: -0.0001

Data Index: 1227923 (C4 (Web Text))

Max Activating Token Index: 81

Text #1

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.4041. Min Act: -0.0001

Data Index: 1378834 (Python Code)

Max Activating Token Index: 581

Text #2

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3671. Min Act: -0.0001

Data Index: 1400451 (Python Code)

Max Activating Token Index: 48

Text #3

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3478. Min Act: -0.0001

Data Index: 1247168 (C4 (Web Text))

Max Activating Token Index: 414

Text #4

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3289. Min Act: -0.0001

Data Index: 268825 (C4 (Web Text))

Max Activating Token Index: 42

Text #5

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3358. Min Act: -0.0001

Data Index: 1145129 (C4 (Web Text))

Max Activating Token Index: 907

Text #6

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3342. Min Act: -0.0001

Data Index: 926166 (C4 (Web Text))

Max Activating Token Index: 29

Text #7

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3210. Min Act: -0.0001

Data Index: 1565175 (Python Code)

Max Activating Token Index: 142

Text #8

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.2760. Min Act: -0.0001

Data Index: 395527 (C4 (Web Text))

Max Activating Token Index: 743

Text #9

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3330. Min Act: -0.0001

Data Index: 163128 (C4 (Web Text))

Max Activating Token Index: 38

Text #10

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.2911. Min Act: -0.0001

Data Index: 1436795 (Python Code)

Max Activating Token Index: 780

Text #11

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3123. Min Act: -0.0001

Data Index: 1091865 (C4 (Web Text))

Max Activating Token Index: 6

Text #12

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3109. Min Act: -0.0001

Data Index: 1461094 (Python Code)

Max Activating Token Index: 457

Text #13

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3410. Min Act: -0.0001

Data Index: 1526160 (Python Code)

Max Activating Token Index: 134

Text #14

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3096. Min Act: -0.0001

Data Index: 929058 (C4 (Web Text))

Max Activating Token Index: 13

Text #15

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3019. Min Act: -0.0001

Data Index: 1391960 (Python Code)

Max Activating Token Index: 181

Text #16

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3250. Min Act: -0.0001

Data Index: 854184 (C4 (Web Text))

Max Activating Token Index: 253

Text #17

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.2938. Min Act: -0.0001

Data Index: 1258771 (C4 (Web Text))

Max Activating Token Index: 3

Text #18

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.2911. Min Act: -0.0001

Data Index: 1649936 (Python Code)

Max Activating Token Index: 365

Text #19

Max Range: 0.4041. Min Range: -0.4041

Max Act: 0.3112. Min Act: -0.0001

Data Index: 756722 (C4 (Web Text))

Max Activating Token Index: 8