Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 2854 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3621. Min Act: -0.0001

Data Index: 410836 (C4 (Web Text))

Max Activating Token Index: 744

Text #1

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3738. Min Act: 0.0000

Data Index: 1322231 (C4 (Web Text))

Max Activating Token Index: 236

Text #2

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3529. Min Act: -0.0001

Data Index: 785104 (C4 (Web Text))

Max Activating Token Index: 873

Text #3

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3411. Min Act: -0.0001

Data Index: 278241 (C4 (Web Text))

Max Activating Token Index: 21

Text #4

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3287. Min Act: -0.0000

Data Index: 376089 (C4 (Web Text))

Max Activating Token Index: 358

Text #5

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3348. Min Act: -0.0001

Data Index: 131676 (C4 (Web Text))

Max Activating Token Index: 792

Text #6

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3320. Min Act: -0.0001

Data Index: 140584 (C4 (Web Text))

Max Activating Token Index: 391

Text #7

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3275. Min Act: -0.0001

Data Index: 429950 (C4 (Web Text))

Max Activating Token Index: 970

Text #8

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3259. Min Act: -0.0001

Data Index: 1233294 (C4 (Web Text))

Max Activating Token Index: 318

Text #9

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3206. Min Act: 0.0000

Data Index: 421835 (C4 (Web Text))

Max Activating Token Index: 652

Text #10

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3287. Min Act: -0.0001

Data Index: 1589033 (Python Code)

Max Activating Token Index: 440

Text #11

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3290. Min Act: -0.0001

Data Index: 196782 (C4 (Web Text))

Max Activating Token Index: 274

Text #12

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3166. Min Act: -0.0001

Data Index: 1595616 (Python Code)

Max Activating Token Index: 850

Text #13

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3134. Min Act: 0.0000

Data Index: 56700 (C4 (Web Text))

Max Activating Token Index: 692

Text #14

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3026. Min Act: 0.0000

Data Index: 1338212 (C4 (Web Text))

Max Activating Token Index: 490

Text #15

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.3114. Min Act: -0.0001

Data Index: 441220 (C4 (Web Text))

Max Activating Token Index: 136

Text #16

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.2995. Min Act: 0.0000

Data Index: 1327899 (C4 (Web Text))

Max Activating Token Index: 231

Text #17

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.2992. Min Act: -0.0001

Data Index: 1296171 (C4 (Web Text))

Max Activating Token Index: 416

Text #18

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.2968. Min Act: -0.0001

Data Index: 156869 (C4 (Web Text))

Max Activating Token Index: 676

Text #19

Max Range: 0.3738. Min Range: -0.3738

Max Act: 0.2942. Min Act: -0.0001

Data Index: 1427204 (Python Code)

Max Activating Token Index: 67