INDEX

Explanations

mixed content/languages

np_max-act · gemini-2.0-flash

The neuron strongly activates on parentheses—especially the “(” token (and to a lesser extent “)” and the end‐of‐text marker)—i.e. it’s detecting parenthetical asides.

oai_token-act-pair · o4-mini

Triggered by @xinyanhu8

responses that advocate for respectful communication regarding weight and body image.

oai_token-act-pair · gpt-4o-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

Negative Logits

 marc

-0.07

.ent

-0.06

 Webcam

-0.06

session

-0.06

(on

-0.06

XL

-0.06

 đồ

-0.06

хови

-0.06

=models

-0.06

.element

-0.06

POSITIVE LOGITS

Japan

0.07

สาม

0.06

 अख

0.06

ÜR

0.06

 postpon

0.06

.Undef

0.06

ouden

0.06

ساس

0.06

(char

0.06

Pokemon

0.06

Activations Density 0.012%

mixed content/languages

The neuron strongly activates on parentheses—especially the “(” token (and to a lesser extent “)” and the end‐of‐text marker)—i.e. it’s detecting parenthetical asides.

responses that advocate for respectful communication regarding weight and body image.

No Known Activations