INDEX

Explanations

references to outdoor activities or settings

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

onymy

-0.55

 présidenti

-0.52

 ainfi

-0.50

 Berlín

-0.50

 Volkes

-0.49

 técn

-0.49

 carottes

-0.48

 Inscrivez

-0.48

 Ansch

-0.47

 pingente

-0.47

POSITIVE LOGITS

 Outdoor

1.09

Outdoor

1.06

 outdoors

1.04

 outdoor

1.02

outdoor

0.98

 Outdoors

0.90

 Outside

0.79

Outside

0.78

 OUTSIDE

0.77

 exterior

0.75

Activations Density 0.002%

references to outdoor activities or settings

No Comments

No Known Activations