Model: GELU Model: 4 Layers, 2048 Neurons per Layer
Dataset: 80% C4 (Web Text) and 20% Python Code
Neuron 2018 in Layer 2
Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')
Text #0
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.7727. Min Act: -0.1700
Data Index: 39048 (C4 (Web Text))
Max Activating Token Index: 936
Click toggle to see full text
Truncated
Full Text #0
Text #1
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.5540. Min Act: -0.1700
Data Index: 62075 (C4 (Web Text))
Max Activating Token Index: 640
Click toggle to see full text
Truncated
Full Text #1
Text #2
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.3977. Min Act: -0.1700
Data Index: 1352395 (C4 (Web Text))
Max Activating Token Index: 498
Click toggle to see full text
Truncated
Full Text #2
Text #3
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.2415. Min Act: -0.1700
Data Index: 499983 (C4 (Web Text))
Max Activating Token Index: 420
Click toggle to see full text
Truncated
Full Text #3
Text #4
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.2415. Min Act: -0.1700
Data Index: 767091 (C4 (Web Text))
Max Activating Token Index: 619
Click toggle to see full text
Truncated
Full Text #4
Text #5
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.2727. Min Act: -0.1700
Data Index: 997542 (C4 (Web Text))
Max Activating Token Index: 619
Click toggle to see full text
Truncated
Full Text #5
Text #6
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.2102. Min Act: -0.1700
Data Index: 647811 (C4 (Web Text))
Max Activating Token Index: 14
Click toggle to see full text
Truncated
Full Text #6
Text #7
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1164. Min Act: -0.1700
Data Index: 407854 (C4 (Web Text))
Max Activating Token Index: 652
Click toggle to see full text
Truncated
Full Text #7
Text #8
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.0852. Min Act: -0.1700
Data Index: 404987 (C4 (Web Text))
Max Activating Token Index: 383
Click toggle to see full text
Truncated
Full Text #8
Text #9
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1477. Min Act: -0.1700
Data Index: 1138910 (C4 (Web Text))
Max Activating Token Index: 523
Click toggle to see full text
Truncated
Full Text #9
Text #10
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1477. Min Act: -0.1700
Data Index: 430151 (C4 (Web Text))
Max Activating Token Index: 111
Click toggle to see full text
Truncated
Full Text #10
Text #11
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1477. Min Act: -0.1700
Data Index: 729914 (C4 (Web Text))
Max Activating Token Index: 862
Click toggle to see full text
Truncated
Full Text #11
Text #12
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1789. Min Act: -0.1700
Data Index: 894731 (C4 (Web Text))
Max Activating Token Index: 396
Click toggle to see full text
Truncated
Full Text #12
Text #13
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.2415. Min Act: -0.1700
Data Index: 790543 (C4 (Web Text))
Max Activating Token Index: 653
Click toggle to see full text
Truncated
Full Text #13
Text #14
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1477. Min Act: -0.1700
Data Index: 965068 (C4 (Web Text))
Max Activating Token Index: 72
Click toggle to see full text
Truncated
Full Text #14
Text #15
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.0852. Min Act: -0.1700
Data Index: 403073 (C4 (Web Text))
Max Activating Token Index: 491
Click toggle to see full text
Truncated
Full Text #15
Text #16
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1164. Min Act: -0.1700
Data Index: 1121301 (C4 (Web Text))
Max Activating Token Index: 93
Click toggle to see full text
Truncated
Full Text #16
Text #17
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1164. Min Act: -0.1700
Data Index: 115629 (C4 (Web Text))
Max Activating Token Index: 899
Click toggle to see full text
Truncated
Full Text #17
Text #18
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.1789. Min Act: -0.1700
Data Index: 712322 (C4 (Web Text))
Max Activating Token Index: 641
Click toggle to see full text
Truncated
Full Text #18
Text #19
Max Range: 4.7727. Min Range: -4.7727
Max Act: 4.0852. Min Act: -0.1700
Data Index: 1264585 (C4 (Web Text))
Max Activating Token Index: 995
Click toggle to see full text
Truncated
Full Text #19