INDEX

Explanations

code/markup

np_max-act · gemini-2.0-flash

The neuron is triggered by metadata‐style or configuration directives (e.g. markup header lines, "#+"-prefix settings, colons, pragma options, HTML_HEAD/TITLE tags, etc.) rather than ordinary prose or code tokens.

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

량

-0.07

 TypeInfo

-0.07

�

-0.07

 Cous

-0.06

條

-0.06

<Token

-0.06

.every

-0.06

rbrace

-0.06

 از

-0.06

 fontsize

-0.06

POSITIVE LOGITS

zeit

0.07

 precisely

0.06

though

0.06

agog

0.06

 Stripe

0.06

_workflow

0.06

CHED

0.06

 حف

0.06

áme

0.06

_MISS

0.06

Activations Density 0.022%

code/markup

The neuron is triggered by metadata‐style or configuration directives (e.g. markup header lines, "#+"-prefix settings, colons, pragma options, HTML_HEAD/TITLE tags, etc.) rather than ordinary prose or code tokens.

No Comments

No Known Activations