INDEX

Explanations

The neuron fires on the key topic nouns that people enthusiastically call out—especially names of animals or giveaway items that users say they “love” or want to see.

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

 stockbild

1.07

 पहला

0.80

...),

0.73

?),

0.72

:"",

0.72

),”

0.71

),"

0.68

ﬁ

0.67

ស្ថានភាព

0.66

*****",

0.66

POSITIVE LOGITS

↵↵↵

3.25

↵↵↵↵

2.97

<eos>

2.91

↵↵

2.91

↵↵↵↵↵

2.80

↵↵↵↵↵↵

2.53

↵↵↵↵↵↵↵

2.47

↵↵↵↵↵↵↵↵

2.23

↵↵↵↵↵↵↵↵↵

2.23

↵↵↵↵↵↵↵↵↵↵↵↵↵

2.23

Activations Density 2.861%