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

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

Neuron 1255 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: 3.9373. Min Range: -3.9373

Max Act: 3.9373. Min Act: -0.1700

Data Index: 1374286 (Python Code)

Max Activating Token Index: 812

Text #1

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.7965. Min Act: -0.1700

Data Index: 457264 (C4 (Web Text))

Max Activating Token Index: 307

Text #2

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.7496. Min Act: -0.1700

Data Index: 1153695 (C4 (Web Text))

Max Activating Token Index: 583

Text #3

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.7809. Min Act: -0.1700

Data Index: 1567348 (Python Code)

Max Activating Token Index: 703

Text #4

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.7027. Min Act: -0.1700

Data Index: 948430 (C4 (Web Text))

Max Activating Token Index: 266

Text #5

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.7027. Min Act: -0.1700

Data Index: 208562 (C4 (Web Text))

Max Activating Token Index: 862

Text #6

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6870. Min Act: -0.1700

Data Index: 1522770 (Python Code)

Max Activating Token Index: 234

Text #7

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.7183. Min Act: -0.1700

Data Index: 1395317 (Python Code)

Max Activating Token Index: 619

Text #8

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.7027. Min Act: -0.1700

Data Index: 595169 (C4 (Web Text))

Max Activating Token Index: 453

Text #9

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6870. Min Act: -0.1700

Data Index: 849792 (C4 (Web Text))

Max Activating Token Index: 358

Text #10

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6401. Min Act: -0.1700

Data Index: 1621397 (Python Code)

Max Activating Token Index: 461

Text #11

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6714. Min Act: -0.1700

Data Index: 1569019 (Python Code)

Max Activating Token Index: 256

Text #12

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6557. Min Act: -0.1700

Data Index: 717909 (C4 (Web Text))

Max Activating Token Index: 1019

Text #13

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6714. Min Act: -0.1700

Data Index: 190428 (C4 (Web Text))

Max Activating Token Index: 449

Text #14

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6714. Min Act: -0.1700

Data Index: 981181 (C4 (Web Text))

Max Activating Token Index: 458

Text #15

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6401. Min Act: -0.1700

Data Index: 1047262 (C4 (Web Text))

Max Activating Token Index: 824

Text #16

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.5931. Min Act: -0.1700

Data Index: 1651969 (Python Code)

Max Activating Token Index: 181

Text #17

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6401. Min Act: -0.1699

Data Index: 1237230 (C4 (Web Text))

Max Activating Token Index: 682

Text #18

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6401. Min Act: -0.1700

Data Index: 1177050 (C4 (Web Text))

Max Activating Token Index: 728

Text #19

Max Range: 3.9373. Min Range: -3.9373

Max Act: 3.6557. Min Act: -0.1700

Data Index: 1404124 (Python Code)

Max Activating Token Index: 700