INDEX

Explanations

capitalized abbreviations in a specific format

oai_token-act-pair · gpt-3.5-turbo

abbreviations or acronyms

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

awed

-0.87

ifications

-0.83

aw

-0.77

hers

-0.74

icles

-0.73

abl

-0.71

nam

-0.70

stood

-0.68

Cly

-0.66

rill

-0.66

POSITIVE LOGITS

LLOW

1.35

LL

1.33

VER

1.29

VE

1.29

ISE

1.28

FILE

1.25

OPER

1.25

KE

1.21

OSE

1.21

OT

1.20

Activations Density 0.062%

capitalized abbreviations in a specific format

abbreviations or acronyms

No Comments

No Known Activations