INDEX

Explanations

I have

The neuron responds to subjective evaluative language in user reviews—especially first‐person pronouns (“I,” “Used”) and sentiment‐laden adjectives or adverbs (e.g. “great,” “perfect,” “suck”).

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

 सत्यापित

-1.49

artamento

-1.36

倏

-1.35

 landes

-1.34

 Daß

-1.33

to

-1.32

人了

-1.29



-1.28

 більш

-1.28

ственном

-1.27

POSITIVE LOGITS

 that

1.99

</tr>

1.59

 your

1.55

利率

1.39

 стихотво

1.38

梨

1.32

 also

1.31

("}");

1.30

 OGSÅ

1.29

幼儿

1.27

Activations Density 0.071%