INDEX

Explanations

The neuron responds strongly to very high‐frequency/common tokens (i.e. words like “of,” “to,” “the,” etc.).

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

<u>

0.64

"],

0.60

⟤

0.58

!!!!!

0.57

Ꮨ

0.57

ᴿ

0.54

—-

0.54

<unused2140>

0.53

!!!!!!

0.52

 😀

0.52

POSITIVE LOGITS

 […]

3.02

 [...]

2.50

 继续访问

2.30

[…]

2.28

 阅读全文

2.25

<start_of_image>

2.19

[...]

2.09

<eos>

2.08

,...

1.82

 (...)

1.77

Activations Density 4.458%