Model: GELU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 1239 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.5223. Min Act: -0.1700

Data Index: 750780 (C4 (Web Text))

Max Activating Token Index: 714

Text #1

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.3812. Min Act: -0.1700

Data Index: 233568 (C4 (Web Text))

Max Activating Token Index: 149

Text #2

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.3185. Min Act: -0.1700

Data Index: 905248 (C4 (Web Text))

Max Activating Token Index: 392

Text #3

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.3342. Min Act: -0.1700

Data Index: 1064721 (C4 (Web Text))

Max Activating Token Index: 66

Text #4

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.2713. Min Act: -0.1700

Data Index: 910399 (C4 (Web Text))

Max Activating Token Index: 639

Text #5

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.2399. Min Act: -0.1700

Data Index: 1494311 (Python Code)

Max Activating Token Index: 206

Text #6

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.2556. Min Act: -0.1700

Data Index: 1146020 (C4 (Web Text))

Max Activating Token Index: 483

Text #7

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.2399. Min Act: -0.1700

Data Index: 944315 (C4 (Web Text))

Max Activating Token Index: 947

Text #8

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.2242. Min Act: -0.1700

Data Index: 1310780 (C4 (Web Text))

Max Activating Token Index: 907

Text #9

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1927. Min Act: -0.1700

Data Index: 766756 (C4 (Web Text))

Max Activating Token Index: 356

Text #10

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.2084. Min Act: -0.1700

Data Index: 1382808 (Python Code)

Max Activating Token Index: 542

Text #11

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1927. Min Act: -0.1700

Data Index: 949245 (C4 (Web Text))

Max Activating Token Index: 348

Text #12

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1927. Min Act: -0.1700

Data Index: 472177 (C4 (Web Text))

Max Activating Token Index: 137

Text #13

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1612. Min Act: -0.1700

Data Index: 53056 (C4 (Web Text))

Max Activating Token Index: 494

Text #14

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1612. Min Act: -0.1700

Data Index: 1395874 (Python Code)

Max Activating Token Index: 683

Text #15

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1770. Min Act: -0.1700

Data Index: 560611 (C4 (Web Text))

Max Activating Token Index: 800

Text #16

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1140. Min Act: -0.1700

Data Index: 1587121 (Python Code)

Max Activating Token Index: 686

Text #17

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1140. Min Act: -0.1700

Data Index: 1222380 (C4 (Web Text))

Max Activating Token Index: 16

Text #18

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1455. Min Act: -0.1700

Data Index: 959457 (C4 (Web Text))

Max Activating Token Index: 340

Text #19

Max Range: 3.5223. Min Range: -3.5223

Max Act: 3.1455. Min Act: -0.1700

Data Index: 96790 (C4 (Web Text))

Max Activating Token Index: 812