INDEX

Explanations

code setup and definitions

The neuron primarily detects comment lines or comment delimiters (e.g. “//” and decorative “====…” separators) in source code.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

it

-1.56

-1.55

-1.45

-1.44

-1.41

 that

-1.40

肏

-1.32

 aislados

-1.32

 nuvem

-1.29

POSITIVE LOGITS

Crítica

1.91

Elaboración

1.89

 plupart

1.80

'},

1.67

 nostre

1.64

 hjälpa

1.59

Parece

1.58

')],

1.55

Hanya

1.54

 terbuat

1.51

Activations Density 0.022%