Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 630 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0065. Min Act: -0.0005

Data Index: 1598623 (Python Code)

Max Activating Token Index: 227

Text #1

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0056. Min Act: -0.0004

Data Index: 1388406 (Python Code)

Max Activating Token Index: 626

Text #2

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0055. Min Act: -0.0005

Data Index: 1373690 (Python Code)

Max Activating Token Index: 196

Text #3

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0054. Min Act: -0.0005

Data Index: 831513 (C4 (Web Text))

Max Activating Token Index: 617

Text #4

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0052. Min Act: -0.0005

Data Index: 1422040 (Python Code)

Max Activating Token Index: 290

Text #5

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0053. Min Act: -0.0005

Data Index: 1521267 (Python Code)

Max Activating Token Index: 610

Text #6

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0052. Min Act: -0.0005

Data Index: 1612860 (Python Code)

Max Activating Token Index: 236

Text #7

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0051. Min Act: -0.0005

Data Index: 451225 (C4 (Web Text))

Max Activating Token Index: 663

Text #8

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0050. Min Act: -0.0006

Data Index: 36288 (C4 (Web Text))

Max Activating Token Index: 242

Text #9

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0050. Min Act: -0.0005

Data Index: 320921 (C4 (Web Text))

Max Activating Token Index: 401

Text #10

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0049. Min Act: -0.0005

Data Index: 412454 (C4 (Web Text))

Max Activating Token Index: 873

Text #11

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0048. Min Act: -0.0005

Data Index: 750377 (C4 (Web Text))

Max Activating Token Index: 509

Text #12

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0047. Min Act: -0.0005

Data Index: 1416494 (Python Code)

Max Activating Token Index: 512

Text #13

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0047. Min Act: -0.0005

Data Index: 1511039 (Python Code)

Max Activating Token Index: 33

Text #14

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0045. Min Act: -0.0005

Data Index: 1194779 (C4 (Web Text))

Max Activating Token Index: 995

Text #15

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0046. Min Act: -0.0005

Data Index: 776458 (C4 (Web Text))

Max Activating Token Index: 614

Text #16

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0046. Min Act: -0.0006

Data Index: 30813 (C4 (Web Text))

Max Activating Token Index: 81

Text #17

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0045. Min Act: -0.0005

Data Index: 1430137 (Python Code)

Max Activating Token Index: 548

Text #18

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0045. Min Act: -0.0005

Data Index: 1437924 (Python Code)

Max Activating Token Index: 698

Text #19

Max Range: 0.0065. Min Range: -0.0065

Max Act: 0.0045. Min Act: -0.0005

Data Index: 518725 (C4 (Web Text))

Max Activating Token Index: 142