INDEX

Explanations

to

np_max-act · gemini-2.0-flash

The neuron detects phrases where the assistant says it’s present to help or assist (e.g. “I’m here to help/assist,” “to help with”).

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

,eg

-0.07

 fileId

-0.07

emiah

-0.06

 Drink

-0.06

 эт

-0.06

 دهند

-0.06

>null

-0.06

 caused

-0.06

 фран

-0.06

<form

-0.06

POSITIVE LOGITS

 ELEMENT

0.07

ホテル

0.07

ané

0.06

(user

0.06

tweet

0.06

 Injury

0.06

BindView

0.06

ValidationError

0.06

.Navigator

0.05

TextField

0.05

Activations Density 0.016%

to

The neuron detects phrases where the assistant says it’s present to help or assist (e.g. “I’m here to help/assist,” “to help with”).

No Comments

No Known Activations