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

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

Neuron 105 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 7.0624. Min Range: -7.0624

Max Act: 7.0624. Min Act: -0.1700

Data Index: 1214700 (C4 (Web Text))

Max Activating Token Index: 943

Text #1

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.9687. Min Act: -0.1700

Data Index: 474831 (C4 (Web Text))

Max Activating Token Index: 197

Text #2

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.9687. Min Act: -0.1700

Data Index: 1544720 (Python Code)

Max Activating Token Index: 892

Text #3

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.8437. Min Act: -0.1700

Data Index: 1645279 (Python Code)

Max Activating Token Index: 443

Text #4

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.8124. Min Act: -0.1700

Data Index: 783816 (C4 (Web Text))

Max Activating Token Index: 51

Text #5

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.7812. Min Act: -0.1700

Data Index: 814451 (C4 (Web Text))

Max Activating Token Index: 525

Text #6

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.7187. Min Act: -0.1700

Data Index: 396724 (C4 (Web Text))

Max Activating Token Index: 228

Text #7

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.6874. Min Act: -0.1700

Data Index: 1585082 (Python Code)

Max Activating Token Index: 851

Text #8

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.8124. Min Act: -0.1700

Data Index: 1367983 (Python Code)

Max Activating Token Index: 897

Text #9

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.5937. Min Act: -0.1700

Data Index: 1566136 (Python Code)

Max Activating Token Index: 678

Text #10

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.7499. Min Act: -0.1700

Data Index: 957942 (C4 (Web Text))

Max Activating Token Index: 974

Text #11

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.7187. Min Act: -0.1700

Data Index: 904613 (C4 (Web Text))

Max Activating Token Index: 378

Text #12

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.6562. Min Act: -0.1700

Data Index: 439344 (C4 (Web Text))

Max Activating Token Index: 67

Text #13

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.7499. Min Act: -0.1700

Data Index: 1544099 (Python Code)

Max Activating Token Index: 580

Text #14

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.5937. Min Act: -0.1700

Data Index: 819006 (C4 (Web Text))

Max Activating Token Index: 241

Text #15

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.6249. Min Act: -0.1700

Data Index: 1341087 (C4 (Web Text))

Max Activating Token Index: 124

Text #16

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.4999. Min Act: -0.1700

Data Index: 1549348 (Python Code)

Max Activating Token Index: 879

Text #17

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.6874. Min Act: -0.1700

Data Index: 457820 (C4 (Web Text))

Max Activating Token Index: 312

Text #18

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.7187. Min Act: -0.1700

Data Index: 393793 (C4 (Web Text))

Max Activating Token Index: 270

Text #19

Max Range: 7.0624. Min Range: -7.0624

Max Act: 6.6249. Min Act: -0.1700

Data Index: 563273 (C4 (Web Text))

Max Activating Token Index: 92