INDEX

Explanations

Security vulnerabilities

np_max-act · gemini-2.0-flash

The neuron activates on terms describing security‐vulnerability exploits—words like “allow,” “gain,” “attacker,” “execute,” “privileges,” etc., that appear in CVE write-ups.

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

ponsive

-0.06

ennis

-0.06

gages

-0.06

ativity

-0.06

nea

-0.06

 Part

-0.06

neas

-0.06

.images

-0.06

separator

-0.06

Unknown

-0.06

POSITIVE LOGITS

 flexGrow

0.07

 Ου

0.07

 Francie

0.07

McG

0.06

.drawLine

0.06

Χ

0.06

vm

0.06

 EINVAL

0.06

Nh

0.06

Endpoints

0.06

Activations Density 0.012%

Security vulnerabilities

The neuron activates on terms describing security‐vulnerability exploits—words like “allow,” “gain,” “attacker,” “execute,” “privileges,” etc., that appear in CVE write-ups.

No Comments

No Known Activations

Security vulnerabilities

The neuron activates on terms describing security‐vulnerability exploits—words like “allow,” “gain,” “attacker,” “execute,” “privileges,” etc., that appear in CVE write-ups.

No Comments

No Known Activations