INDEX

Explanations

references to things that are cherished or valued

oai_token-act-pair · gpt-3.5-turbo

the term "beloved" to indicate affection or fondness toward people, animals, or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ramid

-0.75

annis

-0.72

agnetic

-0.70

enei

-0.70

degree

-0.68

urer

-0.67

err

-0.66

agher

-0.64

omach

-0.64

ulhu

-0.64

POSITIVE LOGITS

 dearly

0.83

 beloved

0.75

 cherished

0.73

 pets

0.66

 darling

0.66

lihood

0.64

endor

0.64

 uncond

0.64

 loved

0.63

 Creatures

0.62

Activations Density 0.015%

references to things that are cherished or valued

the term "beloved" to indicate affection or fondness toward people, animals, or entities

No Comments

No Known Activations