INDEX

Explanations

choosing between options

This neuron identifies words that describe problematic or adverse conditions (e.g. things going wrong) in an advice‐style “if/when” context.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

NOW

0.61

ARGS

0.60

Now

0.60

each

0.60

자들이

0.60

莯

0.60

そして

0.59

CFLAGS

0.59

 jongens

0.59

Targets

0.58

POSITIVE LOGITS

 или

1.58

 அல்லது

1.55

 یا

1.54

 atau

1.53

 hoặc

1.52

 ಅಥವಾ

1.48

 या

1.45

or

1.45

 أو

1.41

 veya

1.41

Activations Density 0.001%