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

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

Neuron 2312 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1172. Min Act: -0.0001

Data Index: 458923 (C4 (Web Text))

Max Activating Token Index: 982

Text #1

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1199. Min Act: -0.0001

Data Index: 1027948 (C4 (Web Text))

Max Activating Token Index: 314

Text #2

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1230. Min Act: -0.0001

Data Index: 1054152 (C4 (Web Text))

Max Activating Token Index: 435

Text #3

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1144. Min Act: -0.0001

Data Index: 295103 (C4 (Web Text))

Max Activating Token Index: 482

Text #4

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1136. Min Act: -0.0001

Data Index: 84921 (C4 (Web Text))

Max Activating Token Index: 230

Text #5

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1270. Min Act: -0.0001

Data Index: 511681 (C4 (Web Text))

Max Activating Token Index: 307

Text #6

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1038. Min Act: -0.0001

Data Index: 864946 (C4 (Web Text))

Max Activating Token Index: 816

Text #7

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1187. Min Act: -0.0001

Data Index: 1233940 (C4 (Web Text))

Max Activating Token Index: 296

Text #8

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1143. Min Act: -0.0001

Data Index: 798614 (C4 (Web Text))

Max Activating Token Index: 424

Text #9

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1165. Min Act: -0.0001

Data Index: 1630299 (Python Code)

Max Activating Token Index: 537

Text #10

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1178. Min Act: -0.0001

Data Index: 989230 (C4 (Web Text))

Max Activating Token Index: 505

Text #11

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.0932. Min Act: -0.0001

Data Index: 6168 (C4 (Web Text))

Max Activating Token Index: 303

Text #12

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1126. Min Act: -0.0001

Data Index: 1291743 (C4 (Web Text))

Max Activating Token Index: 378

Text #13

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.0977. Min Act: -0.0001

Data Index: 1212476 (C4 (Web Text))

Max Activating Token Index: 295

Text #14

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1021. Min Act: -0.0001

Data Index: 1531879 (Python Code)

Max Activating Token Index: 281

Text #15

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1029. Min Act: -0.0001

Data Index: 895069 (C4 (Web Text))

Max Activating Token Index: 426

Text #16

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.1020. Min Act: -0.0001

Data Index: 278217 (C4 (Web Text))

Max Activating Token Index: 203

Text #17

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.0976. Min Act: -0.0001

Data Index: 1620084 (Python Code)

Max Activating Token Index: 321

Text #18

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.0901. Min Act: -0.0001

Data Index: 838598 (C4 (Web Text))

Max Activating Token Index: 429

Text #19

Max Range: 0.1270. Min Range: -0.1270

Max Act: 0.0969. Min Act: -0.0001

Data Index: 327552 (C4 (Web Text))

Max Activating Token Index: 225