INDEX

Explanations

underscores

np_max-act · gemini-2.0-flash

The neuron activates on lines made up of ASCII‐drawing characters (e.g. repeated underscores, slashes, backslashes) and thus detects ASCII art sections.

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

的一个

-0.07

tem

-0.07

_fk

-0.07

!!}

-0.06

_EMP

-0.06

Ap

-0.06

 username

-0.06

струк

-0.06

('.')↵

-0.06

focused

-0.06

POSITIVE LOGITS

orial

0.07

 вас

0.07

 communic

0.06

.begin

0.06

scape

0.06

 Vulner

0.06

 Macedonia

0.06

 ounce

0.06

 symmetric

0.06

<W

0.06

Activations Density 0.009%

underscores

The neuron activates on lines made up of ASCII‐drawing characters (e.g. repeated underscores, slashes, backslashes) and thus detects ASCII art sections.

No Comments

No Known Activations