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

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

Neuron 930 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: 5.5983. Min Range: -5.5983

Max Act: 5.5983. Min Act: -0.1700

Data Index: 589357 (C4 (Web Text))

Max Activating Token Index: 569

Text #1

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.3483. Min Act: -0.1700

Data Index: 85370 (C4 (Web Text))

Max Activating Token Index: 87

Text #2

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.3171. Min Act: -0.1700

Data Index: 859144 (C4 (Web Text))

Max Activating Token Index: 851

Text #3

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.2546. Min Act: -0.1700

Data Index: 1270007 (C4 (Web Text))

Max Activating Token Index: 128

Text #4

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.1296. Min Act: -0.1700

Data Index: 868532 (C4 (Web Text))

Max Activating Token Index: 353

Text #5

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0983. Min Act: -0.1700

Data Index: 202243 (C4 (Web Text))

Max Activating Token Index: 257

Text #6

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.1296. Min Act: -0.1700

Data Index: 341429 (C4 (Web Text))

Max Activating Token Index: 946

Text #7

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0983. Min Act: -0.1700

Data Index: 962177 (C4 (Web Text))

Max Activating Token Index: 274

Text #8

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0046. Min Act: -0.1700

Data Index: 1408878 (Python Code)

Max Activating Token Index: 827

Text #9

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0358. Min Act: -0.1700

Data Index: 1281190 (C4 (Web Text))

Max Activating Token Index: 806

Text #10

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0671. Min Act: -0.1700

Data Index: 780137 (C4 (Web Text))

Max Activating Token Index: 215

Text #11

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0046. Min Act: -0.1700

Data Index: 1134471 (C4 (Web Text))

Max Activating Token Index: 136

Text #12

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0358. Min Act: -0.1700

Data Index: 951211 (C4 (Web Text))

Max Activating Token Index: 189

Text #13

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0046. Min Act: -0.1700

Data Index: 1599787 (Python Code)

Max Activating Token Index: 962

Text #14

Max Range: 5.5983. Min Range: -5.5983

Max Act: 4.9421. Min Act: -0.1700

Data Index: 1217148 (C4 (Web Text))

Max Activating Token Index: 494

Text #15

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0046. Min Act: -0.1700

Data Index: 774910 (C4 (Web Text))

Max Activating Token Index: 285

Text #16

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0046. Min Act: -0.1700

Data Index: 1536884 (Python Code)

Max Activating Token Index: 947

Text #17

Max Range: 5.5983. Min Range: -5.5983

Max Act: 5.0046. Min Act: -0.1700

Data Index: 1287234 (C4 (Web Text))

Max Activating Token Index: 120

Text #18

Max Range: 5.5983. Min Range: -5.5983

Max Act: 4.9733. Min Act: -0.1700

Data Index: 1167770 (C4 (Web Text))

Max Activating Token Index: 152

Text #19

Max Range: 5.5983. Min Range: -5.5983

Max Act: 4.9108. Min Act: -0.1699

Data Index: 27949 (C4 (Web Text))

Max Activating Token Index: 295