INDEX

Explanations

Java list operations and errors

The neuron activates on Java “dot” member‐access or method‐call sequences (identifiers and the “.” in chained calls).

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

 incendie

-0.90

 coucher

-0.85

 svart

-0.82

 Stochastic

-0.81

 PROBLE

-0.81

gob

-0.78

 cubana

-0.77

Efe

-0.77

夜は

-0.77

 cabra

-0.75

POSITIVE LOGITS

stream

1.24

 stream

1.18

forEach

1.00

 streams

0.93

parallel

0.88

]};

0.85

 streaming

0.84

isEmpty

0.79

DispatchToProps

0.79

streamer

0.77

Activations Density 0.006%