INDEX

Explanations

apologies

np_max-act · gemini-2.0-flash

This neuron activates on the assistant’s apology or self‐correction phrases.

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

slot

-0.07

Subscription

-0.07

 Guards

-0.07

gen

-0.06

Coll

-0.06

 Entered

-0.06

charted

-0.06

.userData

-0.06

sec

-0.06

-Mart

-0.06

POSITIVE LOGITS

거

0.06

 hollow

0.06

boys

0.06

 exter

0.06

resh

0.06

aw

0.06

겠

0.06

atr

0.06

 clash

0.06

Vu

0.06

Activations Density 0.016%

apologies

This neuron activates on the assistant’s apology or self‐correction phrases.

No Comments

No Known Activations