INDEX

Explanations

well tolerated

The neuron detects mentions of treatment tolerability (e.g. words like “tolerance,” “tolerated,” and “well tolerated”).

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

</strong>

-2.16

 only

-1.56

 fourth

-1.52

for

-1.49

 remained

-1.42

 that

-1.35

</b>

-1.34

 this

-1.23

 augen

-1.20

聾

-1.20

POSITIVE LOGITS

 Möglichkeit

1.62

 Seperti

1.38

changes

1.34

鹩

1.34

salms

1.32

 occuper

1.31

nels

1.29

 carteles

1.27

nees

1.27

gleichen

1.27

Activations Density 0.037%