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

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

Neuron 224 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.4310. Min Act: -0.0001

Data Index: 15270 (C4 (Web Text))

Max Activating Token Index: 109

Text #1

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.4167. Min Act: -0.0001

Data Index: 242158 (C4 (Web Text))

Max Activating Token Index: 964

Text #2

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3652. Min Act: -0.0001

Data Index: 255055 (C4 (Web Text))

Max Activating Token Index: 192

Text #3

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3611. Min Act: -0.0000

Data Index: 430300 (C4 (Web Text))

Max Activating Token Index: 305

Text #4

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3466. Min Act: -0.0001

Data Index: 1214808 (C4 (Web Text))

Max Activating Token Index: 748

Text #5

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3257. Min Act: -0.0001

Data Index: 602144 (C4 (Web Text))

Max Activating Token Index: 1018

Text #6

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3242. Min Act: -0.0001

Data Index: 55513 (C4 (Web Text))

Max Activating Token Index: 979

Text #7

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3100. Min Act: -0.0001

Data Index: 1531197 (Python Code)

Max Activating Token Index: 479

Text #8

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3032. Min Act: -0.0000

Data Index: 129053 (C4 (Web Text))

Max Activating Token Index: 47

Text #9

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3003. Min Act: -0.0001

Data Index: 577869 (C4 (Web Text))

Max Activating Token Index: 569

Text #10

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2967. Min Act: -0.0001

Data Index: 419654 (C4 (Web Text))

Max Activating Token Index: 1017

Text #11

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2937. Min Act: -0.0000

Data Index: 41034 (C4 (Web Text))

Max Activating Token Index: 685

Text #12

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2940. Min Act: -0.0001

Data Index: 879941 (C4 (Web Text))

Max Activating Token Index: 211

Text #13

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.3009. Min Act: -0.0001

Data Index: 1377062 (Python Code)

Max Activating Token Index: 522

Text #14

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2966. Min Act: -0.0001

Data Index: 1195630 (C4 (Web Text))

Max Activating Token Index: 223

Text #15

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2742. Min Act: -0.0001

Data Index: 1525661 (Python Code)

Max Activating Token Index: 456

Text #16

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2657. Min Act: -0.0001

Data Index: 1513680 (Python Code)

Max Activating Token Index: 843

Text #17

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2629. Min Act: -0.0001

Data Index: 1610266 (Python Code)

Max Activating Token Index: 314

Text #18

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2724. Min Act: -0.0001

Data Index: 1558104 (Python Code)

Max Activating Token Index: 172

Text #19

Max Range: 0.4310. Min Range: -0.4310

Max Act: 0.2667. Min Act: -0.0001

Data Index: 555956 (C4 (Web Text))

Max Activating Token Index: 420