INDEX

Explanations

error logging and reporting

The neuron fires on logging calls—identifiers and method names related to log levels (e.g. Trace, Debug, Info, Warn(ing), Error, Fatal) and print/logger invocations.

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

INFORMATION

-1.00

 INFORMATION

-0.97

 fêtes

-0.94

 green

-0.90

 seeming

-0.87

 pelan

-0.86

твер

-0.86

 חי

-0.86

green

-0.84

 kurang

-0.82

POSITIVE LOGITS

 error

1.21

 horror

1.08

 caught

1.08

 alarmed

1.05

例外

1.02

 warn

0.99

 fatal

0.98

எ

0.98

 warned

0.97

catching

0.96

Activations Density 0.004%