INDEX

Explanations

spdx license comments

The neuron strongly activates on very short tokens (especially two‐character words like “io” or “by”).

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

for

-2.20

 where

-1.77

 these

-1.50

The

-1.49

 they

-1.48

 just

-1.41

 Since

-1.40

 their

-1.37

by

-1.36

 from

-1.34

POSITIVE LOGITS

 resistir

1.52

ciful

1.43

这下

1.39

遑

1.39

 élas

1.38

 sỉ

1.38

又能

1.38

又不

1.35

 той

1.35

为什么会

1.34

Activations Density 0.184%