INDEX

Explanations

confirmation of understanding

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

<unused2113>

1.97

<unused2140>

1.95

AtPath

1.89

гийн

1.85

ických

1.84

䣫

1.84

䂺

1.83

WithObjects

1.83

㫙

1.83

ittä

1.79

POSITIVE LOGITS

4.66

。

3.51

3.12

3.11

).

3.01

↵↵

2.92

।

2.87

2.86

__.

2.83

].

2.81

Activations Density 0.486%