INDEX

Explanations

square brackets in code

The neuron activates on square‐bracket tokens (e.g. “[”, “]”, “[]”, “\[”, “\]”) and thus is looking for bracket delimiters.

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

The

-2.34

These

-1.99

One

-1.96

For

-1.90

and

-1.88

With

-1.85

After

-1.84

 Kasih

-1.83

းက

-1.82

We

-1.80

POSITIVE LOGITS

犼

2.33

at

2.31

2.17

the

2.16

釤

2.11

 inception

2.05

 ファー

2.00

</strong>

1.98

 hvis

1.98

哪裡

1.95

Activations Density 0.265%