INDEX

Explanations

'use strict' patterns

np_acts-logits-general · gemini-2.5-flash-lite

opening brackets or markup symbols that begin structured sections like code blocks, citations, or metadata fields.

oai_token-act-pair · claude-4-5-sonnet

Triggered by @bfgcn7frf5

The neuron strongly activates on numeric tokens (including integers, decimals, and bracketed reference numbers).

oai_token-act-pair · o4-mini

Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_10/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 There

-1.55

mlı

-1.40

 there

-1.40

場合には

-1.38

^(@

-1.36

かどうか

-1.36

かず

-1.35

臊

-1.35

foni

-1.35

これを

-1.33

POSITIVE LOGITS

</h3>

1.76

 {}));

1.71

1.69

 nostro

1.54

1.45

1.44

 spécialiste

1.41

ized

1.39

ial

1.38

Activations Density 0.179%

'use strict' patterns

opening brackets or markup symbols that begin structured sections like code blocks, citations, or metadata fields.

The neuron strongly activates on numeric tokens (including integers, decimals, and bracketed reference numbers).

No Known Activations

'use strict' patterns

opening brackets or markup symbols that begin structured sections like code blocks, citations, or metadata fields.

The neuron strongly activates on numeric tokens (including integers, decimals, and bracketed reference numbers).

No Known Activations