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

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

Neuron 256 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: 5.2306. Min Range: -5.2306

Max Act: 5.2306. Min Act: -0.1700

Data Index: 1161790 (C4 (Web Text))

Max Activating Token Index: 167

Text #1

Max Range: 5.2306. Min Range: -5.2306

Max Act: 5.1369. Min Act: -0.1700

Data Index: 1244865 (C4 (Web Text))

Max Activating Token Index: 240

Text #2

Max Range: 5.2306. Min Range: -5.2306

Max Act: 5.0431. Min Act: -0.1700

Data Index: 570216 (C4 (Web Text))

Max Activating Token Index: 153

Text #3

Max Range: 5.2306. Min Range: -5.2306

Max Act: 5.0744. Min Act: -0.1700

Data Index: 329562 (C4 (Web Text))

Max Activating Token Index: 763

Text #4

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9806. Min Act: -0.1700

Data Index: 76057 (C4 (Web Text))

Max Activating Token Index: 670

Text #5

Max Range: 5.2306. Min Range: -5.2306

Max Act: 5.0119. Min Act: -0.1700

Data Index: 47358 (C4 (Web Text))

Max Activating Token Index: 137

Text #6

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9806. Min Act: -0.1700

Data Index: 295948 (C4 (Web Text))

Max Activating Token Index: 519

Text #7

Max Range: 5.2306. Min Range: -5.2306

Max Act: 5.0119. Min Act: -0.1700

Data Index: 599670 (C4 (Web Text))

Max Activating Token Index: 14

Text #8

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9806. Min Act: -0.1700

Data Index: 276770 (C4 (Web Text))

Max Activating Token Index: 965

Text #9

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9494. Min Act: -0.1700

Data Index: 492409 (C4 (Web Text))

Max Activating Token Index: 449

Text #10

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.8869. Min Act: -0.1700

Data Index: 776679 (C4 (Web Text))

Max Activating Token Index: 35

Text #11

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9181. Min Act: -0.1700

Data Index: 1223816 (C4 (Web Text))

Max Activating Token Index: 20

Text #12

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9494. Min Act: -0.1700

Data Index: 878480 (C4 (Web Text))

Max Activating Token Index: 746

Text #13

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9181. Min Act: -0.1700

Data Index: 912578 (C4 (Web Text))

Max Activating Token Index: 215

Text #14

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.8869. Min Act: -0.1700

Data Index: 550412 (C4 (Web Text))

Max Activating Token Index: 249

Text #15

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.9181. Min Act: -0.1700

Data Index: 487255 (C4 (Web Text))

Max Activating Token Index: 365

Text #16

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.8869. Min Act: -0.1700

Data Index: 341216 (C4 (Web Text))

Max Activating Token Index: 169

Text #17

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.8244. Min Act: -0.1700

Data Index: 1164868 (C4 (Web Text))

Max Activating Token Index: 305

Text #18

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.8869. Min Act: -0.1700

Data Index: 1108430 (C4 (Web Text))

Max Activating Token Index: 405

Text #19

Max Range: 5.2306. Min Range: -5.2306

Max Act: 4.8244. Min Act: -0.1700

Data Index: 320275 (C4 (Web Text))

Max Activating Token Index: 215