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

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

Neuron 84 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.9286. Min Act: -0.1699

Data Index: 1379459 (Python Code)

Max Activating Token Index: 1023

Text #1

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.8261. Min Act: -0.1700

Data Index: 195722 (C4 (Web Text))

Max Activating Token Index: 481

Text #2

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.8433. Min Act: -0.1700

Data Index: 1428480 (Python Code)

Max Activating Token Index: 1023

Text #3

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.8089. Min Act: -0.1700

Data Index: 1575233 (Python Code)

Max Activating Token Index: 1023

Text #4

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.8089. Min Act: -0.1700

Data Index: 678753 (C4 (Web Text))

Max Activating Token Index: 540

Text #5

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7573. Min Act: -0.1700

Data Index: 982765 (C4 (Web Text))

Max Activating Token Index: 724

Text #6

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7573. Min Act: -0.1700

Data Index: 39308 (C4 (Web Text))

Max Activating Token Index: 429

Text #7

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7227. Min Act: -0.1700

Data Index: 776526 (C4 (Web Text))

Max Activating Token Index: 591

Text #8

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7400. Min Act: -0.1700

Data Index: 1397718 (Python Code)

Max Activating Token Index: 1023

Text #9

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7400. Min Act: -0.1700

Data Index: 979172 (C4 (Web Text))

Max Activating Token Index: 982

Text #10

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7227. Min Act: -0.1700

Data Index: 924722 (C4 (Web Text))

Max Activating Token Index: 487

Text #11

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7400. Min Act: -0.1700

Data Index: 1014974 (C4 (Web Text))

Max Activating Token Index: 345

Text #12

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7227. Min Act: -0.1700

Data Index: 19038 (C4 (Web Text))

Max Activating Token Index: 528

Text #13

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7054. Min Act: -0.1698

Data Index: 129672 (C4 (Web Text))

Max Activating Token Index: 537

Text #14

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7227. Min Act: -0.1700

Data Index: 1556877 (Python Code)

Max Activating Token Index: 1023

Text #15

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7227. Min Act: -0.1700

Data Index: 1554690 (Python Code)

Max Activating Token Index: 1023

Text #16

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7054. Min Act: -0.1700

Data Index: 297078 (C4 (Web Text))

Max Activating Token Index: 700

Text #17

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7054. Min Act: -0.1700

Data Index: 67687 (C4 (Web Text))

Max Activating Token Index: 425

Text #18

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7227. Min Act: -0.1699

Data Index: 1471821 (Python Code)

Max Activating Token Index: 1023

Text #19

Max Range: 1.9286. Min Range: -1.9286

Max Act: 1.7054. Min Act: -0.1700

Data Index: 275531 (C4 (Web Text))

Max Activating Token Index: 42