INDEX

Explanations

Respectful and responsible behavior

np_max-act · gemini-2.0-flash

The neuron activates on self-descriptive politeness tokens—especially the word “respectful”—marking polite assistant tone statements.

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

]['

-0.07

.Brand

-0.06

 onboard

-0.06

Managed

-0.06

 cloning

-0.06

Sending

-0.06

\Entities

-0.06

.Score

-0.06

osaurs

-0.06

 prevents

-0.06

POSITIVE LOGITS

 blitz

0.07

expiry

0.06

+-

0.06

parms

0.06

引

0.06

promotion

0.06

 etter

0.06

 využí

0.06

 ödül

0.06

(bin

0.06

Activations Density 0.006%

Respectful and responsible behavior

The neuron activates on self-descriptive politeness tokens—especially the word “respectful”—marking polite assistant tone statements.

No Comments

No Known Activations