Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3540 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.7099. Min Act: -0.0001

Data Index: 1201266 (C4 (Web Text))

Max Activating Token Index: 306

Text #1

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.6161. Min Act: -0.0001

Data Index: 1292078 (C4 (Web Text))

Max Activating Token Index: 655

Text #2

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.6164. Min Act: -0.0001

Data Index: 1142691 (C4 (Web Text))

Max Activating Token Index: 201

Text #3

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5711. Min Act: -0.0001

Data Index: 1096900 (C4 (Web Text))

Max Activating Token Index: 296

Text #4

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5176. Min Act: -0.0001

Data Index: 638748 (C4 (Web Text))

Max Activating Token Index: 893

Text #5

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5290. Min Act: -0.0001

Data Index: 1227286 (C4 (Web Text))

Max Activating Token Index: 277

Text #6

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5446. Min Act: -0.0001

Data Index: 1039994 (C4 (Web Text))

Max Activating Token Index: 729

Text #7

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5189. Min Act: -0.0001

Data Index: 207251 (C4 (Web Text))

Max Activating Token Index: 216

Text #8

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5481. Min Act: -0.0001

Data Index: 957269 (C4 (Web Text))

Max Activating Token Index: 10

Text #9

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5825. Min Act: -0.0001

Data Index: 1318953 (C4 (Web Text))

Max Activating Token Index: 808

Text #10

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5413. Min Act: -0.0001

Data Index: 1113099 (C4 (Web Text))

Max Activating Token Index: 27

Text #11

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5235. Min Act: -0.0001

Data Index: 1541640 (Python Code)

Max Activating Token Index: 99

Text #12

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.4953. Min Act: -0.0001

Data Index: 247662 (C4 (Web Text))

Max Activating Token Index: 584

Text #13

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5087. Min Act: -0.0001

Data Index: 581446 (C4 (Web Text))

Max Activating Token Index: 489

Text #14

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5072. Min Act: -0.0001

Data Index: 244265 (C4 (Web Text))

Max Activating Token Index: 169

Text #15

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5207. Min Act: -0.0001

Data Index: 446362 (C4 (Web Text))

Max Activating Token Index: 737

Text #16

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.4852. Min Act: -0.0001

Data Index: 691808 (C4 (Web Text))

Max Activating Token Index: 242

Text #17

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5131. Min Act: -0.0001

Data Index: 1249395 (C4 (Web Text))

Max Activating Token Index: 31

Text #18

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.4983. Min Act: -0.0001

Data Index: 1157992 (C4 (Web Text))

Max Activating Token Index: 39

Text #19

Max Range: 0.7099. Min Range: -0.7099

Max Act: 0.5290. Min Act: -0.0001

Data Index: 422502 (C4 (Web Text))

Max Activating Token Index: 28