INDEX

Explanations

sentence endings

The neuron fires almost exclusively on the final token of each coherent paragraph or block of text, effectively marking paragraph (or section) boundaries.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 Notably

2.13

 Although

2.06

 Additionally

2.06

 Meanwhile

2.02

 Ironically

1.98

 Furthermore

1.95

 While

1.93

 Despite

1.91

 Interestingly

1.87

 However

1.85

POSITIVE LOGITS

 your

1.41

 type

1.32

 yourself

1.28

1.27

1.26

1.25

1.24

 corrupted

1.23

or

1.21

use

1.21

Activations Density 0.000%