INDEX

Explanations

phrases indicating methods or approaches

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 dezelve

-0.50

prefixer

-0.47

bildēt

-0.46

 Infórmanos

-0.45

 ویکی‌پدی

-0.44

 doméstico

-0.42

ksikon

-0.42

 plegable

-0.41

がいる

-0.40

 nocturno

-0.40

POSITIVE LOGITS

way

0.85

 ways

0.77

Way

0.75

WAY

0.73

 Ways

0.71

 manner

0.70

Way

0.69

 WAYS

0.67

 Manner

0.66

 образом

0.64

Activations Density 0.022%

phrases indicating methods or approaches

No Comments

No Known Activations