INDEX

Explanations

body text

np_max-act · gemini-2.0-flash

This neuron detects meta‐instruction or rubric language referring to the article evaluation (labels like “content,” “article,” “headline,” “delivers,” “provides,” etc.).

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

-0.08

-0.07

 Stand

-0.07

 retired

-0.07

たち

-0.07

))

-0.06

POSITIVE LOGITS

+='<

0.07

.’↵↵

0.07

yay

0.06

 Doug

0.06

?>">
↵

0.06

INSERT

0.06

.",
↵

0.06

 digitally

0.06

Doug

0.06

	private

0.06

Activations Density 0.021%

body text

This neuron detects meta‐instruction or rubric language referring to the article evaluation (labels like “content,” “article,” “headline,” “delivers,” “provides,” etc.).

No Comments

No Known Activations

body text

This neuron detects meta‐instruction or rubric language referring to the article evaluation (labels like “content,” “article,” “headline,” “delivers,” “provides,” etc.).

No Comments

No Known Activations