INDEX

Explanations

Negation

np_max-act · gemini-2.0-flash

The neuron consistently activates on the “’t” suffix in contractions (e.g. don’t), i.e. it detects negation endings.

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

zug

-0.07

out

-0.07

dí

-0.06

 فتح

-0.06

 Ко

-0.06

 debe

-0.06

amed

-0.06

 throne

-0.06

 इस

-0.06

OC

-0.06

POSITIVE LOGITS

[input

0.07

 divers

0.06

 guts

0.06

[k

0.06

 mRNA

0.06

neys

0.06

(parameter

0.06

COOKIE

0.06

(Build

0.06

 UIGraphics

0.06

Activations Density 0.001%

Negation

The neuron consistently activates on the “’t” suffix in contractions (e.g. don’t), i.e. it detects negation endings.

No Comments

No Known Activations