INDEX

Explanations

a

np_max-act · gemini-2.0-flash

The neuron fires on domain‐specific jargon and named‐entity tokens—things like acronyms, product or gene names, specialized technical terms, and other proper nouns.

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

"When

-0.08

 пацієн

-0.07

 errno

-0.07

 یعنی

-0.07

ูกค

-0.07

 escaping

-0.07

)?;↵

-0.07

 représ

-0.06

 GLuint

-0.06

 tiene

-0.06

POSITIVE LOGITS

igated

0.07

鎮

0.07

enen

0.06

цю

0.06

ΑΤ

0.06

cly

0.06

ัล

0.06

ант

0.06

ğinde

0.06

Activations Density 0.137%

a

The neuron fires on domain‐specific jargon and named‐entity tokens—things like acronyms, product or gene names, specialized technical terms, and other proper nouns.

No Comments

No Known Activations