Model: GELU Model: 4 Layers, 2048 Neurons per Layer
Dataset: 80% C4 (Web Text) and 20% Python Code
Neuron 2000 in Layer 2
Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')
Text #0
Max Range: 5.3043. Min Range: -5.3043
Max Act: 5.3043. Min Act: -0.1700
Data Index: 578456 (C4 (Web Text))
Max Activating Token Index: 166
Click toggle to see full text
Truncated
Full Text #0
Text #1
Max Range: 5.3043. Min Range: -5.3043
Max Act: 5.2418. Min Act: -0.1700
Data Index: 161950 (C4 (Web Text))
Max Activating Token Index: 517
Click toggle to see full text
Truncated
Full Text #1
Text #2
Max Range: 5.3043. Min Range: -5.3043
Max Act: 5.1168. Min Act: -0.1700
Data Index: 1212140 (C4 (Web Text))
Max Activating Token Index: 702
Click toggle to see full text
Truncated
Full Text #2
Text #3
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.8980. Min Act: -0.1700
Data Index: 585104 (C4 (Web Text))
Max Activating Token Index: 589
Click toggle to see full text
Truncated
Full Text #3
Text #4
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9293. Min Act: -0.1700
Data Index: 158293 (C4 (Web Text))
Max Activating Token Index: 301
Click toggle to see full text
Truncated
Full Text #4
Text #5
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9293. Min Act: -0.1700
Data Index: 1185292 (C4 (Web Text))
Max Activating Token Index: 449
Click toggle to see full text
Truncated
Full Text #5
Text #6
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9293. Min Act: -0.1700
Data Index: 1049734 (C4 (Web Text))
Max Activating Token Index: 680
Click toggle to see full text
Truncated
Full Text #6
Text #7
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9918. Min Act: -0.1700
Data Index: 671574 (C4 (Web Text))
Max Activating Token Index: 875
Click toggle to see full text
Truncated
Full Text #7
Text #8
Max Range: 5.3043. Min Range: -5.3043
Max Act: 5.0230. Min Act: -0.1700
Data Index: 943533 (C4 (Web Text))
Max Activating Token Index: 794
Click toggle to see full text
Truncated
Full Text #8
Text #9
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.8355. Min Act: -0.1700
Data Index: 117977 (C4 (Web Text))
Max Activating Token Index: 969
Click toggle to see full text
Truncated
Full Text #9
Text #10
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.8668. Min Act: -0.1700
Data Index: 447047 (C4 (Web Text))
Max Activating Token Index: 534
Click toggle to see full text
Truncated
Full Text #10
Text #11
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9918. Min Act: -0.1699
Data Index: 1020131 (C4 (Web Text))
Max Activating Token Index: 220
Click toggle to see full text
Truncated
Full Text #11
Text #12
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9293. Min Act: -0.1700
Data Index: 993573 (C4 (Web Text))
Max Activating Token Index: 717
Click toggle to see full text
Truncated
Full Text #12
Text #13
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9918. Min Act: -0.1700
Data Index: 1174143 (C4 (Web Text))
Max Activating Token Index: 524
Click toggle to see full text
Truncated
Full Text #13
Text #14
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.7418. Min Act: -0.1700
Data Index: 211746 (C4 (Web Text))
Max Activating Token Index: 396
Click toggle to see full text
Truncated
Full Text #14
Text #15
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.8980. Min Act: -0.1700
Data Index: 283049 (C4 (Web Text))
Max Activating Token Index: 337
Click toggle to see full text
Truncated
Full Text #15
Text #16
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9293. Min Act: -0.1700
Data Index: 720668 (C4 (Web Text))
Max Activating Token Index: 692
Click toggle to see full text
Truncated
Full Text #16
Text #17
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.9293. Min Act: -0.1700
Data Index: 783822 (C4 (Web Text))
Max Activating Token Index: 186
Click toggle to see full text
Truncated
Full Text #17
Text #18
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.8668. Min Act: -0.1700
Data Index: 219217 (C4 (Web Text))
Max Activating Token Index: 633
Click toggle to see full text
Truncated
Full Text #18
Text #19
Max Range: 5.3043. Min Range: -5.3043
Max Act: 4.8668. Min Act: -0.1700
Data Index: 1125025 (C4 (Web Text))
Max Activating Token Index: 961
Click toggle to see full text
Truncated
Full Text #19