Model: GELU Model: 4 Layers, 2048 Neurons per Layer
Dataset: 80% C4 (Web Text) and 20% Python Code
Neuron 1984 in Layer 1
Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')
Text #0
Max Range: 4.1573. Min Range: -4.1573
Max Act: 4.1573. Min Act: -0.1700
Data Index: 735654 (C4 (Web Text))
Max Activating Token Index: 184
Click toggle to see full text
Truncated
Full Text #0
Text #1
Max Range: 4.1573. Min Range: -4.1573
Max Act: 4.0948. Min Act: -0.1700
Data Index: 773101 (C4 (Web Text))
Max Activating Token Index: 779
Click toggle to see full text
Truncated
Full Text #1
Text #2
Max Range: 4.1573. Min Range: -4.1573
Max Act: 4.0322. Min Act: -0.1700
Data Index: 1643061 (Python Code)
Max Activating Token Index: 647
Click toggle to see full text
Truncated
Full Text #2
Text #3
Max Range: 4.1573. Min Range: -4.1573
Max Act: 4.0010. Min Act: -0.1700
Data Index: 398789 (C4 (Web Text))
Max Activating Token Index: 735
Click toggle to see full text
Truncated
Full Text #3
Text #4
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.9697. Min Act: -0.1700
Data Index: 120936 (C4 (Web Text))
Max Activating Token Index: 259
Click toggle to see full text
Truncated
Full Text #4
Text #5
Max Range: 4.1573. Min Range: -4.1573
Max Act: 4.0010. Min Act: -0.1700
Data Index: 1271019 (C4 (Web Text))
Max Activating Token Index: 405
Click toggle to see full text
Truncated
Full Text #5
Text #6
Max Range: 4.1573. Min Range: -4.1573
Max Act: 4.0322. Min Act: -0.1700
Data Index: 486244 (C4 (Web Text))
Max Activating Token Index: 96
Click toggle to see full text
Truncated
Full Text #6
Text #7
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.9697. Min Act: -0.1700
Data Index: 1421080 (Python Code)
Max Activating Token Index: 377
Click toggle to see full text
Truncated
Full Text #7
Text #8
Max Range: 4.1573. Min Range: -4.1573
Max Act: 4.0010. Min Act: -0.1700
Data Index: 1193467 (C4 (Web Text))
Max Activating Token Index: 719
Click toggle to see full text
Truncated
Full Text #8
Text #9
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.9384. Min Act: -0.1700
Data Index: 1338486 (C4 (Web Text))
Max Activating Token Index: 364
Click toggle to see full text
Truncated
Full Text #9
Text #10
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.9384. Min Act: -0.1700
Data Index: 663957 (C4 (Web Text))
Max Activating Token Index: 270
Click toggle to see full text
Truncated
Full Text #10
Text #11
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.9072. Min Act: -0.1700
Data Index: 651867 (C4 (Web Text))
Max Activating Token Index: 514
Click toggle to see full text
Truncated
Full Text #11
Text #12
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.9072. Min Act: -0.1700
Data Index: 222057 (C4 (Web Text))
Max Activating Token Index: 800
Click toggle to see full text
Truncated
Full Text #12
Text #13
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.8759. Min Act: -0.1700
Data Index: 1165584 (C4 (Web Text))
Max Activating Token Index: 484
Click toggle to see full text
Truncated
Full Text #13
Text #14
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.8759. Min Act: -0.1700
Data Index: 1132532 (C4 (Web Text))
Max Activating Token Index: 589
Click toggle to see full text
Truncated
Full Text #14
Text #15
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.8759. Min Act: -0.1700
Data Index: 672807 (C4 (Web Text))
Max Activating Token Index: 469
Click toggle to see full text
Truncated
Full Text #15
Text #16
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.8446. Min Act: -0.1700
Data Index: 819396 (C4 (Web Text))
Max Activating Token Index: 268
Click toggle to see full text
Truncated
Full Text #16
Text #17
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.8446. Min Act: -0.1700
Data Index: 1144105 (C4 (Web Text))
Max Activating Token Index: 974
Click toggle to see full text
Truncated
Full Text #17
Text #18
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.8446. Min Act: -0.1700
Data Index: 1526271 (Python Code)
Max Activating Token Index: 906
Click toggle to see full text
Truncated
Full Text #18
Text #19
Max Range: 4.1573. Min Range: -4.1573
Max Act: 3.8446. Min Act: -0.1700
Data Index: 541351 (C4 (Web Text))
Max Activating Token Index: 370
Click toggle to see full text
Truncated
Full Text #19