INDEX

Explanations

code

np_max-act · gemini-2.0-flash

The neuron activates on negative or harmful descriptive terms—e.g. toxic, poisonous, harmful, malevolent, deleterious—flagging words that convey malicious or damaging intent.

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

#ga

-0.07

_nombre

-0.06

 ApplicationContext

-0.06

 Bernardino

-0.06

,L

-0.06

logradouro

-0.06

allah

-0.06

ndata

-0.06

caling

-0.06

illisecond

-0.06

POSITIVE LOGITS

 Advertising

0.07

 sensit

0.07

TOM

0.07

>To

0.07

 listopadu

0.06

Recovered

0.06

 SHOULD

0.06

 güncel

0.06

 presenting

0.06

:";↵

0.06

Activations Density 0.006%

code

The neuron activates on negative or harmful descriptive terms—e.g. toxic, poisonous, harmful, malevolent, deleterious—flagging words that convey malicious or damaging intent.

No Comments

No Known Activations